Add application file

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+third_party/ml-depth-pro/data/example.jpg filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -0,0 +1,2 @@
+*.pth
+*.pt

app.py

@@ -0,0 +1,172 @@
+# Copyright (C) 2024-present Naver Corporation. All rights reserved.
+# Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+#
+# --------------------------------------------------------
+# gradio demo
+# --------------------------------------------------------
+
+import argparse
+import math
+import gradio
+import os
+import torch
+import numpy as np
+import tempfile
+import functools
+import copy
+from tqdm import tqdm
+import cv2
+
+from dust3r.inference import inference
+from dust3r.model import AsymmetricCroCo3DStereo
+from dust3r.image_pairs import make_pairs
+from dust3r.utils.image_pose import load_images, rgb, enlarge_seg_masks
+from dust3r.utils.device import to_numpy
+from dust3r.cloud_opt_flow import global_aligner, GlobalAlignerMode
+import matplotlib.pyplot as pl
+from transformers import pipeline
+from dust3r.utils.viz_demo import convert_scene_output_to_glb
+import depth_pro
+pl.ion()
+
+# for gpu >= Ampere and pytorch >= 1.12
+torch.backends.cuda.matmul.allow_tf32 = True
+batch_size = 1
+
+tmpdirname = tempfile.mkdtemp(suffix='_align3r_gradio_demo')
+image_size = 512
+silent = True
+gradio_delete_cache = 7200
+
+
+class FileState:
+    def __init__(self, outfile_name=None):
+        self.outfile_name = outfile_name
+
+    def __del__(self):
+        if self.outfile_name is not None and os.path.isfile(self.outfile_name):
+            os.remove(self.outfile_name)
+        self.outfile_name = None
+
+def get_3D_model_from_scene(outdir, silent, scene, min_conf_thr=3, as_pointcloud=False, mask_sky=False,
+                            clean_depth=False, transparent_cams=False, cam_size=0.05, show_cam=True, save_name=None, thr_for_init_conf=True):
+    """
+    extract 3D_model (glb file) from a reconstructed scene
+    """
+    if scene is None:
+        return None
+    # post processes
+    if clean_depth:
+        scene = scene.clean_pointcloud()
+    if mask_sky:
+        scene = scene.mask_sky()
+
+    # get optimized values from scene
+    rgbimg = scene.imgs
+    focals = scene.get_focals().cpu()
+    cams2world = scene.get_im_poses().cpu()
+    # 3D pointcloud from depthmap, poses and intrinsics
+    pts3d = to_numpy(scene.get_pts3d(raw_pts=True))
+    scene.min_conf_thr = min_conf_thr
+    scene.thr_for_init_conf = thr_for_init_conf
+    msk = to_numpy(scene.get_masks())
+    cmap = pl.get_cmap('viridis')
+    cam_color = [cmap(i/len(rgbimg))[:3] for i in range(len(rgbimg))]
+    cam_color = [(255*c[0], 255*c[1], 255*c[2]) for c in cam_color]
+    return convert_scene_output_to_glb(outdir, rgbimg, pts3d, msk, focals, cams2world, as_pointcloud=as_pointcloud,
+                                        transparent_cams=transparent_cams, cam_size=cam_size, show_cam=show_cam, silent=silent, save_name=save_name,
+                                        cam_color=cam_color)
+
+def generate_monocular_depth_maps(img_list, depth_prior_name):
+    if depth_prior_name=='depthpro':
+        model, transform = depth_pro.create_model_and_transforms(device='cuda')
+        model.eval()
+        for image_path in tqdm(img_list):
+          path_depthpro = image_path.replace('.png','_pred_depth_depthpro.npz').replace('.jpg','_pred_depth_depthpro.npz')
+          image, _, f_px = depth_pro.load_rgb(image_path)
+          image = transform(image)
+          # Run inference.
+          prediction = model.infer(image, f_px=f_px)
+          depth = prediction["depth"].cpu()  # Depth in [m].
+          np.savez_compressed(path_depthpro, depth=depth, focallength_px=prediction["focallength_px"].cpu())  
+    elif depth_prior_name=='depthanything':
+        pipe = pipeline(task="depth-estimation", model="depth-anything/Depth-Anything-V2-Large-hf",device='cuda')
+        for image_path in tqdm(img_list):
+          path_depthanything = image_path.replace('.png','_pred_depth_depthanything.npz').replace('.jpg','_pred_depth_depthanything.npz')
+          image = Image.open(image_path)
+          depth = pipe(image)["predicted_depth"].numpy()
+          np.savez_compressed(path_depthanything, depth=depth)  
+
+@spaces.GPU(duration=180)
+def local_get_reconstructed_scene(filelist, min_conf_thr, as_pointcloud, mask_sky, clean_depth, transparent_cams, cam_size, depth_prior_name, **kw):
+    generate_monocular_depth_maps(filelist, depth_prior_name)
+    imgs = load_images(filelist, size=image_size, verbose=not silent,traj_format='custom', depth_prior_name=depth_prior_name)
+    pairs = []
+    pairs.append((imgs[0], imgs[1]))
+    output = inference(pairs, model, device, batch_size=batch_size, verbose=not silent)
+    mode = GlobalAlignerMode.PairViewer
+    scene = global_aligner(output, device=device, mode=mode, verbose=not silent)
+    save_folder = './output'
+    outfile = get_3D_model_from_scene(save_folder, silent, scene, min_conf_thr, as_pointcloud, mask_sky, clean_depth, transparent_cams, cam_size, show_cam)
+      
+    return outfile
+
+
+def run_example(snapshot, matching_conf_thr, min_conf_thr, cam_size, as_pointcloud, shared_intrinsics, filelist, **kw):
+    return local_get_reconstructed_scene(filelist, cam_size, **kw)
+
+css = """.gradio-container {margin: 0 !important; min-width: 100%};"""
+title = "Align3R Demo"
+with gradio.Blocks(css=css, title=title, delete_cache=(gradio_delete_cache, gradio_delete_cache)) as demo:
+    filestate = gradio.State(None)
+    gradio.HTML('<h2 style="text-align: center;">3D Reconstruction with MASt3R</h2>')
+    gradio.HTML('<p>Upload two images (wait for them to be fully uploaded before hitting the run button). '
+                'If you want to try larger image collections, you can find the more complete version of this demo that you can run locally '
+                'and more details about the method at <a href="https://github.com/jiah-cloud/Align3R">github.com/jiah-cloud/Align3R</a>. '
+                'The checkpoint used in this demo is available at <a href="https://huggingface.co/cyun9286/Align3R_DepthAnythingV2_ViTLarge_BaseDecoder_512_dpt">Align3R (Depth Anything V2)</a> and <a href="https://huggingface.co/cyun9286/Align3R_DepthPro_ViTLarge_BaseDecoder_512_dpt">Align3R (Depth Pro)</a>.</p>')
+    with gradio.Column():
+        inputfiles = gradio.File(file_count="multiple")
+        snapshot = gradio.Image(None, visible=False)
+        with gradio.Row():
+            # adjust the camera size in the output pointcloud
+            cam_size = gradio.Slider(label="cam_size", value=0.2, minimum=0.001, maximum=1.0, step=0.001)
+
+            depth_prior_name = gradio.Dropdown(
+            ["Depth Pro", "Depth Anything V2"], label="monocular depth estimation model", info="Select the monocular depth estimation model.")
+            min_conf_thr = gradio.Slider(label="min_conf_thr", value=1.1, minimum=0.0, maximum=20, step=0.01)
+
+            if depth_prior_name == "Depth Pro":
+              weights_path = "cyun9286/Align3R_DepthPro_ViTLarge_BaseDecoder_512_dpt"
+            else:
+              weights_path = "cyun9286/Align3R_DepthAnythingV2_ViTLarge_BaseDecoder_512_dpt"
+            device = 'cuda' if torch.cuda.is_available() else 'cpu'
+            model = AsymmetricCroCo3DStereo.from_pretrained(weights_path).to(device)  
+        with gradio.Row():
+            as_pointcloud = gradio.Checkbox(value=True, label="As pointcloud")
+            mask_sky = gradio.Checkbox(value=False, label="Mask sky")
+            clean_depth = gradio.Checkbox(value=True, label="Clean-up depthmaps")
+            transparent_cams = gradio.Checkbox(value=False, label="Transparent cameras")
+            # not to show camera
+            show_cam = gradio.Checkbox(value=True, label="Show Camera")
+        run_btn = gradio.Button("Run")
+        outmodel = gradio.Model3D()
+
+        # examples = gradio.Examples(
+        #     examples=[
+        #         ['./example/yellowman/frame_0003.png',
+        #             0.0, 1.5, 0.2, True, False,
+        #         ]
+        #     ],
+        #     inputs=[snapshot, matching_conf_thr, min_conf_thr, cam_size, as_pointcloud, shared_intrinsics, inputfiles],
+        #     outputs=[filestate, outmodel],
+        #     fn=run_example,
+        #     cache_examples="lazy",
+        # )
+
+        # events
+        run_btn.click(fn=local_get_reconstructed_scene,
+                      inputs=[inputfiles, min_conf_thr, as_pointcloud, mask_sky, clean_depth, transparent_cams, cam_size, depth_prior_name],
+                      outputs=[outmodel])
+
+demo.launch(show_error=True, share=None, server_name=None, server_port=None)
+shutil.rmtree(tmpdirname)

croco/LICENSE

@@ -0,0 +1,52 @@
+CroCo, Copyright (c) 2022-present Naver Corporation, is licensed under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 license.
+
+A summary of the CC BY-NC-SA 4.0 license is located here:
+	https://creativecommons.org/licenses/by-nc-sa/4.0/
+
+The CC BY-NC-SA 4.0 license is located here:
+	https://creativecommons.org/licenses/by-nc-sa/4.0/legalcode
+	
+	
+SEE NOTICE BELOW WITH RESPECT TO THE FILE: models/pos_embed.py, models/blocks.py
+
+***************************
+
+NOTICE WITH RESPECT TO THE FILE: models/pos_embed.py
+
+This software is being redistributed in a modifiled form. The original form is available here:
+
+https://github.com/facebookresearch/mae/blob/main/util/pos_embed.py
+
+This software in this file incorporates parts of the following software available here:
+
+Transformer: https://github.com/tensorflow/models/blob/master/official/legacy/transformer/model_utils.py
+available under the following license: https://github.com/tensorflow/models/blob/master/LICENSE
+
+MoCo v3: https://github.com/facebookresearch/moco-v3
+available under the following license: https://github.com/facebookresearch/moco-v3/blob/main/LICENSE
+
+DeiT: https://github.com/facebookresearch/deit
+available under the following license: https://github.com/facebookresearch/deit/blob/main/LICENSE
+
+
+ORIGINAL COPYRIGHT NOTICE AND PERMISSION NOTICE AVAILABLE HERE IS REPRODUCE BELOW:
+
+https://github.com/facebookresearch/mae/blob/main/LICENSE
+
+Attribution-NonCommercial 4.0 International
+
+***************************
+
+NOTICE WITH RESPECT TO THE FILE: models/blocks.py
+
+This software is being redistributed in a modifiled form. The original form is available here:
+
+https://github.com/rwightman/pytorch-image-models
+
+ORIGINAL COPYRIGHT NOTICE AND PERMISSION NOTICE AVAILABLE HERE IS REPRODUCE BELOW:
+
+https://github.com/rwightman/pytorch-image-models/blob/master/LICENSE
+
+Apache License
+Version 2.0, January 2004
+http://www.apache.org/licenses/

croco/NOTICE

@@ -0,0 +1,21 @@
+CroCo
+Copyright 2022-present NAVER Corp.
+
+This project contains subcomponents with separate copyright notices and license terms. 
+Your use of the source code for these subcomponents is subject to the terms and conditions of the following licenses.
+
+====
+
+facebookresearch/mae
+https://github.com/facebookresearch/mae
+
+Attribution-NonCommercial 4.0 International
+
+====
+
+rwightman/pytorch-image-models
+https://github.com/rwightman/pytorch-image-models
+
+Apache License
+Version 2.0, January 2004
+http://www.apache.org/licenses/

croco/README.MD

@@ -0,0 +1,124 @@
+# CroCo + CroCo v2 / CroCo-Stereo / CroCo-Flow
+
+[[`CroCo arXiv`](https://arxiv.org/abs/2210.10716)] [[`CroCo v2 arXiv`](https://arxiv.org/abs/2211.10408)] [[`project page and demo`](https://croco.europe.naverlabs.com/)]
+
+This repository contains the code for our CroCo model presented in our NeurIPS'22 paper [CroCo: Self-Supervised Pre-training for 3D Vision Tasks by Cross-View Completion](https://openreview.net/pdf?id=wZEfHUM5ri) and its follow-up extension published at ICCV'23 [Improved Cross-view Completion Pre-training for Stereo Matching and Optical Flow](https://openaccess.thecvf.com/content/ICCV2023/html/Weinzaepfel_CroCo_v2_Improved_Cross-view_Completion_Pre-training_for_Stereo_Matching_and_ICCV_2023_paper.html), refered to as CroCo v2:
+
+![image](assets/arch.jpg)
+
+```bibtex
+@inproceedings{croco,
+  title={{CroCo: Self-Supervised Pre-training for 3D Vision Tasks by Cross-View Completion}},
+  author={{Weinzaepfel, Philippe and Leroy, Vincent and Lucas, Thomas and Br\'egier, Romain and Cabon, Yohann and Arora, Vaibhav and Antsfeld, Leonid and Chidlovskii, Boris and Csurka, Gabriela and Revaud J\'er\^ome}},
+  booktitle={{NeurIPS}},
+  year={2022}
+}
+
+@inproceedings{croco_v2,
+  title={{CroCo v2: Improved Cross-view Completion Pre-training for Stereo Matching and Optical Flow}},
+  author={Weinzaepfel, Philippe and Lucas, Thomas and Leroy, Vincent and Cabon, Yohann and Arora, Vaibhav and Br{\'e}gier, Romain and Csurka, Gabriela and Antsfeld, Leonid and Chidlovskii, Boris and Revaud, J{\'e}r{\^o}me}, 
+  booktitle={ICCV},
+  year={2023}
+}
+```
+
+## License
+
+The code is distributed under the CC BY-NC-SA 4.0 License. See [LICENSE](LICENSE) for more information.
+Some components are based on code from [MAE](https://github.com/facebookresearch/mae) released under the CC BY-NC-SA 4.0 License and [timm](https://github.com/rwightman/pytorch-image-models) released under the Apache 2.0 License.
+Some components for stereo matching and optical flow are based on code from [unimatch](https://github.com/autonomousvision/unimatch) released under the MIT license.
+
+## Preparation
+
+1. Install dependencies on a machine with a NVidia GPU using e.g. conda. Note that `habitat-sim` is required only for the interactive demo and the synthetic pre-training data generation. If you don't plan to use it, you can ignore the line installing it and use a more recent python version.
+
+```bash
+conda create -n croco python=3.7 cmake=3.14.0
+conda activate croco
+conda install habitat-sim headless -c conda-forge -c aihabitat
+conda install pytorch torchvision -c pytorch
+conda install notebook ipykernel matplotlib
+conda install ipywidgets widgetsnbextension
+conda install scikit-learn tqdm quaternion opencv # only for pretraining / habitat data generation
+
+```
+
+2. Compile cuda kernels for RoPE
+
+CroCo v2 relies on RoPE positional embeddings for which you need to compile some cuda kernels.
+```bash
+cd models/curope/
+python setup.py build_ext --inplace
+cd ../../
+```
+
+This can be a bit long as we compile for all cuda architectures, feel free to update L9 of `models/curope/setup.py` to compile for specific architectures only.
+You might also need to set the environment `CUDA_HOME` in case you use a custom cuda installation.
+
+In case you cannot provide, we also provide a slow pytorch version, which will be automatically loaded.
+
+3. Download pre-trained model
+
+We provide several pre-trained models:
+
+| modelname                                                                                                                          | pre-training data | pos. embed. | Encoder | Decoder |
+|------------------------------------------------------------------------------------------------------------------------------------|-------------------|-------------|---------|---------|
+| [`CroCo.pth`](https://download.europe.naverlabs.com/ComputerVision/CroCo/CroCo.pth)                                                 | Habitat           | cosine      | ViT-B   | Small   |
+| [`CroCo_V2_ViTBase_SmallDecoder.pth`](https://download.europe.naverlabs.com/ComputerVision/CroCo/CroCo_V2_ViTBase_SmallDecoder.pth) | Habitat + real    | RoPE        | ViT-B   | Small   |
+| [`CroCo_V2_ViTBase_BaseDecoder.pth`](https://download.europe.naverlabs.com/ComputerVision/CroCo/CroCo_V2_ViTBase_BaseDecoder.pth)   | Habitat + real    | RoPE        | ViT-B   | Base    |
+| [`CroCo_V2_ViTLarge_BaseDecoder.pth`](https://download.europe.naverlabs.com/ComputerVision/CroCo/CroCo_V2_ViTLarge_BaseDecoder.pth) | Habitat + real    | RoPE        | ViT-L   | Base    |
+
+To download a specific model, i.e., the first one (`CroCo.pth`)
+```bash
+mkdir -p pretrained_models/
+wget https://download.europe.naverlabs.com/ComputerVision/CroCo/CroCo.pth -P pretrained_models/
+```
+
+## Reconstruction example
+
+Simply run after downloading the `CroCo_V2_ViTLarge_BaseDecoder` pretrained model (or update the corresponding line in `demo.py`)
+```bash
+python demo.py
+```
+
+## Interactive demonstration of cross-view completion reconstruction on the Habitat simulator
+
+First download the test scene from Habitat:
+```bash
+python -m habitat_sim.utils.datasets_download --uids habitat_test_scenes --data-path habitat-sim-data/
+```
+
+Then, run the Notebook demo `interactive_demo.ipynb`.
+
+In this demo, you should be able to sample a random reference viewpoint from an [Habitat](https://github.com/facebookresearch/habitat-sim) test scene. Use the sliders to change viewpoint and select a masked target view to reconstruct using CroCo.
+![croco_interactive_demo](https://user-images.githubusercontent.com/1822210/200516576-7937bc6a-55f8-49ed-8618-3ddf89433ea4.jpg)
+
+## Pre-training 
+
+### CroCo 
+
+To pre-train CroCo, please first generate the pre-training data from the Habitat simulator, following the instructions in [datasets/habitat_sim/README.MD](datasets/habitat_sim/README.MD) and then run the following command:
+```
+torchrun --nproc_per_node=4 pretrain.py --output_dir ./output/pretraining/
+```
+
+Our CroCo pre-training was launched on a single server with 4 GPUs.
+It should take around 10 days with A100 or 15 days with V100 to do the 400 pre-training epochs, but decent performances are obtained earlier in training.
+Note that, while the code contains the same scaling rule of the learning rate as MAE when changing the effective batch size, we did not experimented if it is valid in our case.
+The first run can take a few minutes to start, to parse all available pre-training pairs.
+
+### CroCo v2 
+
+For CroCo v2 pre-training, in addition to the generation of the pre-training data from the Habitat simulator above, please pre-extract the crops from the real datasets following the instructions in [datasets/crops/README.MD](datasets/crops/README.MD).
+Then, run the following command for the largest model (ViT-L encoder, Base decoder):
+```
+torchrun --nproc_per_node=8 pretrain.py --model "CroCoNet(enc_embed_dim=1024, enc_depth=24, enc_num_heads=16, dec_embed_dim=768, dec_num_heads=12, dec_depth=12, pos_embed='RoPE100')" --dataset "habitat_release+ARKitScenes+MegaDepth+3DStreetView+IndoorVL" --warmup_epochs 12 --max_epoch 125 --epochs 250 --amp 0 --keep_freq 5 --output_dir ./output/pretraining_crocov2/
+```
+
+Our CroCo v2 pre-training was launched on a single server with 8 GPUs for the largest model, and on a single server with 4 GPUs for the smaller ones, keeping a batch size of 64 per gpu in all cases.
+The largest model should take around 12 days on A100.
+Note that, while the code contains the same scaling rule of the learning rate as MAE when changing the effective batch size, we did not experimented if it is valid in our case.
+
+## Stereo matching and Optical flow downstream tasks
+
+For CroCo-Stereo and CroCo-Flow, please refer to [stereoflow/README.MD](stereoflow/README.MD).

croco/croco-stereo-flow-demo.ipynb

@@ -0,0 +1,191 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "id": "9bca0f41",
+   "metadata": {},
+   "source": [
+    "# Simple inference example with CroCo-Stereo or CroCo-Flow"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "80653ef7",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Copyright (C) 2022-present Naver Corporation. All rights reserved.\n",
+    "# Licensed under CC BY-NC-SA 4.0 (non-commercial use only)."
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "4f033862",
+   "metadata": {},
+   "source": [
+    "First download the model(s) of your choice by running\n",
+    "```\n",
+    "bash stereoflow/download_model.sh crocostereo.pth\n",
+    "bash stereoflow/download_model.sh crocoflow.pth\n",
+    "```"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "1fb2e392",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import torch\n",
+    "use_gpu = torch.cuda.is_available() and torch.cuda.device_count()>0\n",
+    "device = torch.device('cuda:0' if use_gpu else 'cpu')\n",
+    "import matplotlib.pylab as plt"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "e0e25d77",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from stereoflow.test import _load_model_and_criterion\n",
+    "from stereoflow.engine import tiled_pred\n",
+    "from stereoflow.datasets_stereo import img_to_tensor, vis_disparity\n",
+    "from stereoflow.datasets_flow import flowToColor\n",
+    "tile_overlap=0.7 # recommended value, higher value can be slightly better but slower"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "86a921f5",
+   "metadata": {},
+   "source": [
+    "### CroCo-Stereo example"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "64e483cb",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "image1 = np.asarray(Image.open('<path_to_left_image>'))\n",
+    "image2 = np.asarray(Image.open('<path_to_right_image>'))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "f0d04303",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "model, _, cropsize, with_conf, task, tile_conf_mode = _load_model_and_criterion('stereoflow_models/crocostereo.pth', None, device)\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "47dc14b5",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "im1 = img_to_tensor(image1).to(device).unsqueeze(0)\n",
+    "im2 = img_to_tensor(image2).to(device).unsqueeze(0)\n",
+    "with torch.inference_mode():\n",
+    "    pred, _, _ = tiled_pred(model, None, im1, im2, None, conf_mode=tile_conf_mode, overlap=tile_overlap, crop=cropsize, with_conf=with_conf, return_time=False)\n",
+    "pred = pred.squeeze(0).squeeze(0).cpu().numpy()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "583b9f16",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "plt.imshow(vis_disparity(pred))\n",
+    "plt.axis('off')"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "id": "d2df5d70",
+   "metadata": {},
+   "source": [
+    "### CroCo-Flow example"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "9ee257a7",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "image1 = np.asarray(Image.open('<path_to_first_image>'))\n",
+    "image2 = np.asarray(Image.open('<path_to_second_image>'))"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "d5edccf0",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "model, _, cropsize, with_conf, task, tile_conf_mode = _load_model_and_criterion('stereoflow_models/crocoflow.pth', None, device)\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "b19692c3",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "im1 = img_to_tensor(image1).to(device).unsqueeze(0)\n",
+    "im2 = img_to_tensor(image2).to(device).unsqueeze(0)\n",
+    "with torch.inference_mode():\n",
+    "    pred, _, _ = tiled_pred(model, None, im1, im2, None, conf_mode=tile_conf_mode, overlap=tile_overlap, crop=cropsize, with_conf=with_conf, return_time=False)\n",
+    "pred = pred.squeeze(0).permute(1,2,0).cpu().numpy()"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "26f79db3",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "plt.imshow(flowToColor(pred))\n",
+    "plt.axis('off')"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.9.7"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

croco/datasets/crops/README.MD

@@ -0,0 +1,104 @@
+## Generation of crops from the real datasets
+
+The instructions below allow to generate the crops used for pre-training CroCo v2 from the following real-world datasets: ARKitScenes, MegaDepth, 3DStreetView and IndoorVL.
+
+### Download the metadata of the crops to generate 
+
+First, download the metadata and put them in `./data/`:
+```
+mkdir -p data
+cd data/
+wget https://download.europe.naverlabs.com/ComputerVision/CroCo/data/crop_metadata.zip
+unzip crop_metadata.zip
+rm crop_metadata.zip
+cd ..
+```
+
+### Prepare the original datasets 
+
+Second, download the original datasets in `./data/original_datasets/`.
+```
+mkdir -p data/original_datasets
+```
+
+##### ARKitScenes
+
+Download the `raw` dataset from https://github.com/apple/ARKitScenes/blob/main/DATA.md and put it in `./data/original_datasets/ARKitScenes/`.
+The resulting file structure should be like:
+```
+./data/original_datasets/ARKitScenes/
+└───Training
+    └───40753679
+     │  │   ultrawide
+     │  │   ...
+     └───40753686
+     │   
+      ...
+```
+
+##### MegaDepth
+
+Download `MegaDepth v1 Dataset` from https://www.cs.cornell.edu/projects/megadepth/ and put it in `./data/original_datasets/MegaDepth/`.
+The resulting file structure should be like:
+
+```
+./data/original_datasets/MegaDepth/
+└───0000
+│   └───images
+│    │      │   1000557903_87fa96b8a4_o.jpg
+│    │      └ ...
+│    └─── ...
+└───0001
+│   │   
+│   └ ...
+└─── ...
+```
+
+##### 3DStreetView
+
+Download `3D_Street_View` dataset from https://github.com/amir32002/3D_Street_View and put it in `./data/original_datasets/3DStreetView/`.
+The resulting file structure should be like:
+
+``` 
+./data/original_datasets/3DStreetView/
+└───dataset_aligned
+│   └───0002
+│    │      │   0000002_0000001_0000002_0000001.jpg
+│    │      └ ...
+│    └─── ...
+└───dataset_unaligned
+│   └───0003
+│    │      │   0000003_0000001_0000002_0000001.jpg
+│    │      └ ...
+│    └─── ...
+```
+
+##### IndoorVL
+
+Download the `IndoorVL` datasets using [Kapture](https://github.com/naver/kapture).
+
+```
+pip install kapture
+mkdir -p ./data/original_datasets/IndoorVL
+cd ./data/original_datasets/IndoorVL
+kapture_download_dataset.py update
+kapture_download_dataset.py install  "HyundaiDepartmentStore_*"
+kapture_download_dataset.py install  "GangnamStation_*"
+cd -
+```
+
+### Extract the crops
+
+Now, extract the crops for each of the dataset:
+```
+for dataset in ARKitScenes MegaDepth 3DStreetView IndoorVL; 
+do 
+  python3 datasets/crops/extract_crops_from_images.py --crops ./data/crop_metadata/${dataset}/crops_release.txt --root-dir ./data/original_datasets/${dataset}/ --output-dir ./data/${dataset}_crops/ --imsize 256 --nthread 8 --max-subdir-levels 5 --ideal-number-pairs-in-dir 500;
+done
+```
+
+##### Note for IndoorVL
+
+Due to some legal issues, we can only release 144,228 pairs out of the 1,593,689 pairs used in the paper.
+To account for it in terms of number of pre-training iterations, the pre-training command in this repository uses 125 training epochs including 12 warm-up epochs and learning rate cosine schedule of 250, instead of 100, 10 and 200 respectively.
+The impact on the performance is negligible.

croco/datasets/crops/extract_crops_from_images.py

@@ -0,0 +1,159 @@
+# Copyright (C) 2022-present Naver Corporation. All rights reserved.
+# Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+# 
+# --------------------------------------------------------
+# Extracting crops for pre-training
+# --------------------------------------------------------
+
+import os
+import argparse
+from tqdm import tqdm
+from PIL import Image
+import functools
+from multiprocessing import Pool
+import math
+
+
+def arg_parser():
+    parser = argparse.ArgumentParser('Generate cropped image pairs from image crop list')
+
+    parser.add_argument('--crops', type=str, required=True, help='crop file')
+    parser.add_argument('--root-dir', type=str, required=True, help='root directory')
+    parser.add_argument('--output-dir', type=str, required=True, help='output directory')
+    parser.add_argument('--imsize', type=int, default=256, help='size of the crops')
+    parser.add_argument('--nthread', type=int, required=True, help='number of simultaneous threads')
+    parser.add_argument('--max-subdir-levels', type=int, default=5, help='maximum number of subdirectories')
+    parser.add_argument('--ideal-number-pairs-in-dir', type=int, default=500, help='number of pairs stored in a dir')
+    return parser
+
+
+def main(args):
+    listing_path = os.path.join(args.output_dir, 'listing.txt')
+
+    print(f'Loading list of crops ... ({args.nthread} threads)')
+    crops, num_crops_to_generate = load_crop_file(args.crops)
+
+    print(f'Preparing jobs ({len(crops)} candidate image pairs)...')
+    num_levels = min(math.ceil(math.log(num_crops_to_generate, args.ideal_number_pairs_in_dir)), args.max_subdir_levels)
+    num_pairs_in_dir = math.ceil(num_crops_to_generate ** (1/num_levels))
+
+    jobs = prepare_jobs(crops, num_levels, num_pairs_in_dir)
+    del crops
+
+    os.makedirs(args.output_dir, exist_ok=True)
+    mmap = Pool(args.nthread).imap_unordered if args.nthread > 1 else map
+    call = functools.partial(save_image_crops, args)
+
+    print(f"Generating cropped images to {args.output_dir} ...")
+    with open(listing_path, 'w') as listing:
+        listing.write('# pair_path\n')
+        for results in tqdm(mmap(call, jobs), total=len(jobs)):
+            for path in results:
+                listing.write(f'{path}\n')
+    print('Finished writing listing to', listing_path)
+
+
+def load_crop_file(path):
+    data = open(path).read().splitlines()
+    pairs = []
+    num_crops_to_generate = 0
+    for line in tqdm(data):
+        if line.startswith('#'):
+            continue
+        line = line.split(', ')
+        if len(line) < 8:
+            img1, img2, rotation = line
+            pairs.append((img1, img2, int(rotation), []))
+        else:
+            l1, r1, t1, b1, l2, r2, t2, b2 = map(int, line)
+            rect1, rect2 = (l1, t1, r1, b1), (l2, t2, r2, b2)
+            pairs[-1][-1].append((rect1, rect2))
+            num_crops_to_generate += 1
+    return pairs, num_crops_to_generate
+
+
+def prepare_jobs(pairs, num_levels, num_pairs_in_dir):
+    jobs = []
+    powers = [num_pairs_in_dir**level for level in reversed(range(num_levels))]
+
+    def get_path(idx):
+        idx_array = []
+        d = idx
+        for level in range(num_levels - 1):
+            idx_array.append(idx // powers[level])
+            idx = idx % powers[level]
+        idx_array.append(d)
+        return '/'.join(map(lambda x: hex(x)[2:], idx_array))
+
+    idx = 0
+    for pair_data in tqdm(pairs):
+        img1, img2, rotation, crops = pair_data
+        if -60 <= rotation and rotation <= 60:
+            rotation = 0  # most likely not a true rotation
+        paths = [get_path(idx + k) for k in range(len(crops))]
+        idx += len(crops)
+        jobs.append(((img1, img2), rotation, crops, paths))
+    return jobs
+
+
+def load_image(path):
+    try:
+        return Image.open(path).convert('RGB')
+    except Exception as e:
+        print('skipping', path, e)
+        raise OSError()
+
+
+def save_image_crops(args, data):
+    # load images
+    img_pair, rot, crops, paths = data
+    try:
+        img1, img2 = [load_image(os.path.join(args.root_dir, impath)) for impath in img_pair]
+    except OSError as e:
+        return []
+
+    def area(sz):
+        return sz[0] * sz[1]
+
+    tgt_size = (args.imsize, args.imsize)
+
+    def prepare_crop(img, rect, rot=0):
+        # actual crop
+        img = img.crop(rect)
+
+        # resize to desired size
+        interp = Image.Resampling.LANCZOS if area(img.size) > 4*area(tgt_size) else Image.Resampling.BICUBIC
+        img = img.resize(tgt_size, resample=interp)
+
+        # rotate the image
+        rot90 = (round(rot/90) % 4) * 90
+        if rot90 == 90:
+            img = img.transpose(Image.Transpose.ROTATE_90)
+        elif rot90 == 180:
+            img = img.transpose(Image.Transpose.ROTATE_180)
+        elif rot90 == 270:
+            img = img.transpose(Image.Transpose.ROTATE_270)
+        return img
+
+    results = []
+    for (rect1, rect2), path in zip(crops, paths):
+        crop1 = prepare_crop(img1, rect1)
+        crop2 = prepare_crop(img2, rect2, rot)
+
+        fullpath1 = os.path.join(args.output_dir,  path+'_1.jpg')
+        fullpath2 = os.path.join(args.output_dir,  path+'_2.jpg')
+        os.makedirs(os.path.dirname(fullpath1), exist_ok=True)
+
+        assert not os.path.isfile(fullpath1), fullpath1
+        assert not os.path.isfile(fullpath2), fullpath2
+        crop1.save(fullpath1)
+        crop2.save(fullpath2)
+        results.append(path)
+
+    return results
+
+
+if __name__ == '__main__':
+    args = arg_parser().parse_args()
+    main(args)
+

croco/datasets/habitat_sim/README.MD

@@ -0,0 +1,76 @@
+## Generation of synthetic image pairs using Habitat-Sim
+
+These instructions allow to generate pre-training pairs from the Habitat simulator.
+As we did not save metadata of the pairs used in the original paper, they are not strictly the same, but these data use the same setting and are equivalent.
+
+### Download Habitat-Sim scenes
+Download Habitat-Sim scenes:
+- Download links can be found here: https://github.com/facebookresearch/habitat-sim/blob/main/DATASETS.md
+- We used scenes from the HM3D, habitat-test-scenes, Replica, ReplicaCad and ScanNet datasets.
+- Please put the scenes under `./data/habitat-sim-data/scene_datasets/` following the structure below, or update manually paths in `paths.py`.
+```
+./data/
+└──habitat-sim-data/
+   └──scene_datasets/
+      ├──hm3d/
+      ├──gibson/
+      ├──habitat-test-scenes/
+      ├──replica_cad_baked_lighting/
+      ├──replica_cad/
+      ├──ReplicaDataset/
+      └──scannet/
+```
+
+### Image pairs generation
+We provide metadata to generate reproducible images pairs for pretraining and validation.
+Experiments described in the paper used similar data, but whose generation was not reproducible at the time.
+
+Specifications:
+- 256x256 resolution images, with 60 degrees field of view .
+- Up to 1000 image pairs per scene.
+- Number of scenes considered/number of images pairs per dataset:
+  - Scannet: 1097 scenes / 985 209 pairs
+  - HM3D:
+    - hm3d/train: 800 / 800k pairs
+    - hm3d/val: 100 scenes / 100k pairs
+    - hm3d/minival: 10 scenes / 10k pairs
+  - habitat-test-scenes: 3 scenes / 3k pairs
+  - replica_cad_baked_lighting: 13 scenes / 13k pairs
+
+- Scenes from hm3d/val and hm3d/minival pairs were not used for the pre-training but kept for validation purposes.
+
+Download metadata and extract it:
+```bash
+mkdir -p data/habitat_release_metadata/
+cd data/habitat_release_metadata/
+wget https://download.europe.naverlabs.com/ComputerVision/CroCo/data/habitat_release_metadata/multiview_habitat_metadata.tar.gz
+tar -xvf multiview_habitat_metadata.tar.gz
+cd ../..
+# Location of the metadata
+METADATA_DIR="./data/habitat_release_metadata/multiview_habitat_metadata"
+```
+
+Generate image pairs from metadata:
+- The following command will print a list of commandlines to generate image pairs for each scene:
+```bash
+# Target output directory
+PAIRS_DATASET_DIR="./data/habitat_release/"
+python datasets/habitat_sim/generate_from_metadata_files.py --input_dir=$METADATA_DIR --output_dir=$PAIRS_DATASET_DIR
+```
+- One can launch multiple of such commands in parallel e.g. using GNU Parallel:
+```bash
+python datasets/habitat_sim/generate_from_metadata_files.py --input_dir=$METADATA_DIR --output_dir=$PAIRS_DATASET_DIR | parallel -j 16
+```
+
+## Metadata generation
+
+Image pairs were randomly sampled using the following commands, whose outputs contain randomness and are thus not exactly reproducible:
+```bash
+# Print commandlines to generate image pairs from the different scenes available.
+PAIRS_DATASET_DIR=MY_CUSTOM_PATH
+python datasets/habitat_sim/generate_multiview_images.py --list_commands --output_dir=$PAIRS_DATASET_DIR
+
+# Once a dataset is generated, pack metadata files for reproducibility.
+METADATA_DIR=MY_CUSTON_PATH
+python datasets/habitat_sim/pack_metadata_files.py $PAIRS_DATASET_DIR  $METADATA_DIR
+```

croco/datasets/habitat_sim/generate_from_metadata.py

@@ -0,0 +1,92 @@
+# Copyright (C) 2022-present Naver Corporation. All rights reserved.
+# Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+
+"""
+Script to generate image pairs for a given scene reproducing poses provided in a metadata file.
+"""
+import os
+from datasets.habitat_sim.multiview_habitat_sim_generator import MultiviewHabitatSimGenerator
+from datasets.habitat_sim.paths import SCENES_DATASET
+import argparse
+import quaternion
+import PIL.Image
+import cv2
+import json
+from tqdm import tqdm
+
+def generate_multiview_images_from_metadata(metadata_filename,
+                                            output_dir,
+                                            overload_params = dict(),
+                                            scene_datasets_paths=None,
+                                            exist_ok=False):   
+    """
+    Generate images from a metadata file for reproducibility purposes.
+    """
+    # Reorder paths by decreasing label length, to avoid collisions when testing if a string by such label
+    if scene_datasets_paths is not None:
+        scene_datasets_paths = dict(sorted(scene_datasets_paths.items(), key= lambda x: len(x[0]), reverse=True))
+
+    with open(metadata_filename, 'r') as f:
+        input_metadata = json.load(f)
+    metadata = dict()
+    for key, value in input_metadata.items():
+        # Optionally replace some paths
+        if key in ("scene_dataset_config_file", "scene", "navmesh") and value != "":
+            if scene_datasets_paths is not None:
+                for dataset_label, dataset_path in scene_datasets_paths.items():
+                    if value.startswith(dataset_label):
+                        value = os.path.normpath(os.path.join(dataset_path, os.path.relpath(value, dataset_label)))
+                        break
+        metadata[key] = value
+
+    # Overload some parameters
+    for key, value in overload_params.items():
+        metadata[key] = value
+
+    generation_entries = dict([(key, value) for key, value in metadata.items() if not (key in ('multiviews', 'output_dir', 'generate_depth'))])
+    generate_depth = metadata["generate_depth"]
+
+    os.makedirs(output_dir, exist_ok=exist_ok)
+ 
+    generator = MultiviewHabitatSimGenerator(**generation_entries)
+
+    # Generate views
+    for idx_label, data in tqdm(metadata['multiviews'].items()):
+        positions = data["positions"]
+        orientations = data["orientations"]
+        n = len(positions)
+        for oidx in range(n):
+            observation = generator.render_viewpoint(positions[oidx], quaternion.from_float_array(orientations[oidx]))
+            observation_label = f"{oidx + 1}" # Leonid is indexing starting from 1
+            # Color image saved using PIL
+            img = PIL.Image.fromarray(observation['color'][:,:,:3])
+            filename = os.path.join(output_dir, f"{idx_label}_{observation_label}.jpeg")
+            img.save(filename)
+            if generate_depth:
+                # Depth image as EXR file
+                filename = os.path.join(output_dir, f"{idx_label}_{observation_label}_depth.exr")
+                cv2.imwrite(filename, observation['depth'], [cv2.IMWRITE_EXR_TYPE, cv2.IMWRITE_EXR_TYPE_HALF])
+                # Camera parameters
+                camera_params = dict([(key, observation[key].tolist()) for key in ("camera_intrinsics", "R_cam2world", "t_cam2world")])
+                filename = os.path.join(output_dir, f"{idx_label}_{observation_label}_camera_params.json")
+                with open(filename, "w") as f:
+                    json.dump(camera_params, f)
+                # Save metadata
+    with open(os.path.join(output_dir, "metadata.json"), "w") as f:
+        json.dump(metadata, f)
+
+    generator.close()
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--metadata_filename", required=True)
+    parser.add_argument("--output_dir", required=True)
+    args = parser.parse_args()
+
+    generate_multiview_images_from_metadata(metadata_filename=args.metadata_filename,
+                             output_dir=args.output_dir,
+                             scene_datasets_paths=SCENES_DATASET,
+                             overload_params=dict(),
+                             exist_ok=True)
+
+ 

croco/datasets/habitat_sim/generate_from_metadata_files.py

@@ -0,0 +1,27 @@
+# Copyright (C) 2022-present Naver Corporation. All rights reserved.
+# Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+
+"""
+Script generating commandlines to generate image pairs from metadata files.
+"""
+import os
+import glob
+from tqdm import tqdm
+import argparse
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--input_dir", required=True)
+    parser.add_argument("--output_dir", required=True)
+    parser.add_argument("--prefix", default="", help="Commanline prefix, useful e.g. to setup environment.")
+    args = parser.parse_args()
+
+    input_metadata_filenames = glob.iglob(f"{args.input_dir}/**/metadata.json", recursive=True)
+
+    for metadata_filename in tqdm(input_metadata_filenames):
+        output_dir = os.path.join(args.output_dir, os.path.relpath(os.path.dirname(metadata_filename), args.input_dir))
+        # Do not process the scene if the metadata file already exists
+        if os.path.exists(os.path.join(output_dir, "metadata.json")):
+            continue
+        commandline = f"{args.prefix}python datasets/habitat_sim/generate_from_metadata.py --metadata_filename={metadata_filename} --output_dir={output_dir}"
+        print(commandline)

croco/datasets/habitat_sim/generate_multiview_images.py

@@ -0,0 +1,177 @@
+# Copyright (C) 2022-present Naver Corporation. All rights reserved.
+# Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+
+import os
+from tqdm import tqdm
+import argparse
+import PIL.Image
+import numpy as np
+import json
+from datasets.habitat_sim.multiview_habitat_sim_generator import MultiviewHabitatSimGenerator, NoNaviguableSpaceError
+from datasets.habitat_sim.paths import list_scenes_available
+import cv2
+import quaternion
+import shutil
+
+def generate_multiview_images_for_scene(scene_dataset_config_file,
+                                        scene,
+                                        navmesh,
+                                        output_dir, 
+                                        views_count,
+                                        size, 
+                                        exist_ok=False, 
+                                        generate_depth=False,
+                                        **kwargs):
+    """
+    Generate tuples of overlapping views for a given scene.
+    generate_depth: generate depth images and camera parameters.
+    """
+    if os.path.exists(output_dir) and not exist_ok:
+        print(f"Scene {scene}: data already generated. Ignoring generation.")
+        return
+    try:
+        print(f"Scene {scene}: {size} multiview acquisitions to generate...")
+        os.makedirs(output_dir, exist_ok=exist_ok)
+
+        metadata_filename = os.path.join(output_dir, "metadata.json")
+
+        metadata_template = dict(scene_dataset_config_file=scene_dataset_config_file,
+            scene=scene, 
+            navmesh=navmesh,
+            views_count=views_count,
+            size=size,
+            generate_depth=generate_depth,
+            **kwargs)
+        metadata_template["multiviews"] = dict()
+
+        if os.path.exists(metadata_filename):
+            print("Metadata file already exists:", metadata_filename)
+            print("Loading already generated metadata file...")
+            with open(metadata_filename, "r") as f:
+                metadata = json.load(f)
+
+            for key in metadata_template.keys():
+                if key != "multiviews":
+                    assert metadata_template[key] == metadata[key], f"existing file is inconsistent with the input parameters:\nKey: {key}\nmetadata: {metadata[key]}\ntemplate: {metadata_template[key]}."
+        else:
+            print("No temporary file found. Starting generation from scratch...")
+            metadata = metadata_template
+
+        starting_id = len(metadata["multiviews"])
+        print(f"Starting generation from index {starting_id}/{size}...")
+        if starting_id >= size:
+            print("Generation already done.")
+            return
+
+        generator = MultiviewHabitatSimGenerator(scene_dataset_config_file=scene_dataset_config_file,
+                                                scene=scene,
+                                                navmesh=navmesh,
+                                                views_count = views_count,
+                                                size = size,
+                                                **kwargs)
+
+        for idx in tqdm(range(starting_id, size)):
+            # Generate / re-generate the observations
+            try:
+                data = generator[idx]
+                observations = data["observations"]
+                positions = data["positions"]
+                orientations = data["orientations"]
+
+                idx_label = f"{idx:08}"
+                for oidx, observation in enumerate(observations):
+                    observation_label = f"{oidx + 1}" # Leonid is indexing starting from 1
+                    # Color image saved using PIL
+                    img = PIL.Image.fromarray(observation['color'][:,:,:3])
+                    filename = os.path.join(output_dir, f"{idx_label}_{observation_label}.jpeg")
+                    img.save(filename)
+                    if generate_depth:
+                        # Depth image as EXR file
+                        filename = os.path.join(output_dir, f"{idx_label}_{observation_label}_depth.exr")
+                        cv2.imwrite(filename, observation['depth'], [cv2.IMWRITE_EXR_TYPE, cv2.IMWRITE_EXR_TYPE_HALF])
+                        # Camera parameters
+                        camera_params = dict([(key, observation[key].tolist()) for key in ("camera_intrinsics", "R_cam2world", "t_cam2world")])
+                        filename = os.path.join(output_dir, f"{idx_label}_{observation_label}_camera_params.json")
+                        with open(filename, "w") as f:
+                            json.dump(camera_params, f)
+                metadata["multiviews"][idx_label] = {"positions": positions.tolist(),
+                                                    "orientations": orientations.tolist(),
+                                                    "covisibility_ratios": data["covisibility_ratios"].tolist(),
+                                                    "valid_fractions": data["valid_fractions"].tolist(),
+                                                    "pairwise_visibility_ratios": data["pairwise_visibility_ratios"].tolist()}
+            except RecursionError:
+                print("Recursion error: unable to sample observations for this scene. We will stop there.")
+                break
+
+            # Regularly save a temporary metadata file, in case we need to restart the generation
+            if idx % 10 == 0:
+                with open(metadata_filename, "w") as f:
+                    json.dump(metadata, f)
+
+        # Save metadata
+        with open(metadata_filename, "w") as f:
+            json.dump(metadata, f)
+
+        generator.close()
+    except NoNaviguableSpaceError:
+        pass
+
+def create_commandline(scene_data, generate_depth, exist_ok=False):
+    """
+    Create a commandline string to generate a scene.
+    """
+    def my_formatting(val):
+        if val is None or val == "":
+            return '""'
+        else:
+            return val
+    commandline = f"""python {__file__} --scene {my_formatting(scene_data.scene)} 
+    --scene_dataset_config_file {my_formatting(scene_data.scene_dataset_config_file)} 
+    --navmesh {my_formatting(scene_data.navmesh)} 
+    --output_dir {my_formatting(scene_data.output_dir)} 
+    --generate_depth {int(generate_depth)} 
+    --exist_ok {int(exist_ok)}
+    """
+    commandline = " ".join(commandline.split())
+    return commandline
+
+if __name__ == "__main__":
+    os.umask(2)
+
+    parser = argparse.ArgumentParser(description="""Example of use -- listing commands to generate data for scenes available:
+    > python datasets/habitat_sim/generate_multiview_habitat_images.py --list_commands
+    """)
+    
+    parser.add_argument("--output_dir", type=str, required=True)
+    parser.add_argument("--list_commands", action='store_true', help="list commandlines to run if true")
+    parser.add_argument("--scene", type=str, default="")
+    parser.add_argument("--scene_dataset_config_file", type=str, default="")
+    parser.add_argument("--navmesh", type=str, default="")
+    
+    parser.add_argument("--generate_depth", type=int, default=1)
+    parser.add_argument("--exist_ok", type=int, default=0)
+
+    kwargs = dict(resolution=(256,256), hfov=60, views_count = 2, size=1000)
+
+    args = parser.parse_args()
+    generate_depth=bool(args.generate_depth)
+    exist_ok = bool(args.exist_ok)
+
+    if args.list_commands:
+        # Listing scenes available...
+        scenes_data = list_scenes_available(base_output_dir=args.output_dir)
+        
+        for scene_data in scenes_data:
+            print(create_commandline(scene_data, generate_depth=generate_depth, exist_ok=exist_ok))
+    else:
+        if args.scene == "" or args.output_dir == "":
+            print("Missing scene or output dir argument!")
+            print(parser.format_help())
+        else:
+            generate_multiview_images_for_scene(scene=args.scene,
+                                                scene_dataset_config_file = args.scene_dataset_config_file,
+                                                navmesh = args.navmesh,
+                                                output_dir = args.output_dir,
+                                                exist_ok=exist_ok,
+                                                generate_depth=generate_depth,
+                                                **kwargs)

croco/datasets/habitat_sim/multiview_habitat_sim_generator.py

@@ -0,0 +1,390 @@
+# Copyright (C) 2022-present Naver Corporation. All rights reserved.
+# Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+
+import os
+import numpy as np
+import quaternion
+import habitat_sim
+import json
+from sklearn.neighbors import NearestNeighbors
+import cv2
+
+# OpenCV to habitat camera convention transformation
+R_OPENCV2HABITAT = np.stack((habitat_sim.geo.RIGHT, -habitat_sim.geo.UP, habitat_sim.geo.FRONT), axis=0)
+R_HABITAT2OPENCV = R_OPENCV2HABITAT.T
+DEG2RAD = np.pi / 180
+
+def compute_camera_intrinsics(height, width, hfov):
+    f = width/2 / np.tan(hfov/2 * np.pi/180)
+    cu, cv = width/2, height/2
+    return f, cu, cv
+
+def compute_camera_pose_opencv_convention(camera_position, camera_orientation):
+    R_cam2world = quaternion.as_rotation_matrix(camera_orientation) @ R_OPENCV2HABITAT
+    t_cam2world = np.asarray(camera_position)
+    return R_cam2world, t_cam2world
+
+def compute_pointmap(depthmap, hfov):
+    """ Compute a HxWx3 pointmap in camera frame from a HxW depth map."""
+    height, width = depthmap.shape
+    f, cu, cv = compute_camera_intrinsics(height, width, hfov)
+    # Cast depth map to point
+    z_cam = depthmap
+    u, v = np.meshgrid(range(width), range(height))
+    x_cam = (u - cu) / f * z_cam
+    y_cam = (v - cv) / f * z_cam
+    X_cam = np.stack((x_cam, y_cam, z_cam), axis=-1)
+    return X_cam
+
+def compute_pointcloud(depthmap, hfov, camera_position, camera_rotation):
+    """Return a 3D point cloud corresponding to valid pixels of the depth map"""
+    R_cam2world, t_cam2world = compute_camera_pose_opencv_convention(camera_position, camera_rotation)
+
+    X_cam = compute_pointmap(depthmap=depthmap, hfov=hfov)
+    valid_mask = (X_cam[:,:,2] != 0.0)
+
+    X_cam = X_cam.reshape(-1, 3)[valid_mask.flatten()]
+    X_world = X_cam @ R_cam2world.T + t_cam2world.reshape(1, 3)
+    return X_world
+
+def compute_pointcloud_overlaps_scikit(pointcloud1, pointcloud2, distance_threshold, compute_symmetric=False):
+    """
+    Compute 'overlapping' metrics based on a distance threshold between two point clouds.
+    """
+    nbrs = NearestNeighbors(n_neighbors=1, algorithm = 'kd_tree').fit(pointcloud2)
+    distances, indices = nbrs.kneighbors(pointcloud1)
+    intersection1 = np.count_nonzero(distances.flatten() < distance_threshold)
+
+    data = {"intersection1": intersection1,
+            "size1": len(pointcloud1)}
+    if compute_symmetric:
+        nbrs = NearestNeighbors(n_neighbors=1, algorithm = 'kd_tree').fit(pointcloud1)
+        distances, indices = nbrs.kneighbors(pointcloud2)
+        intersection2 = np.count_nonzero(distances.flatten() < distance_threshold)
+        data["intersection2"] = intersection2
+        data["size2"] = len(pointcloud2)
+
+    return data
+
+def _append_camera_parameters(observation, hfov, camera_location, camera_rotation):
+    """
+    Add camera parameters to the observation dictionnary produced by Habitat-Sim
+    In-place modifications.
+    """
+    R_cam2world, t_cam2world = compute_camera_pose_opencv_convention(camera_location, camera_rotation)
+    height, width = observation['depth'].shape
+    f, cu, cv = compute_camera_intrinsics(height, width, hfov)
+    K = np.asarray([[f, 0, cu],
+                    [0, f, cv],
+                    [0, 0, 1.0]])
+    observation["camera_intrinsics"] = K
+    observation["t_cam2world"] = t_cam2world
+    observation["R_cam2world"] = R_cam2world
+
+def look_at(eye, center, up, return_cam2world=True):
+    """
+    Return camera pose looking at a given center point.
+    Analogous of gluLookAt function, using OpenCV camera convention.
+    """
+    z = center - eye
+    z /= np.linalg.norm(z, axis=-1, keepdims=True)
+    y = -up
+    y = y - np.sum(y * z, axis=-1, keepdims=True) * z
+    y /= np.linalg.norm(y, axis=-1, keepdims=True)
+    x = np.cross(y, z, axis=-1)
+
+    if return_cam2world:
+        R = np.stack((x, y, z), axis=-1)
+        t = eye
+    else:
+        # World to camera transformation
+        # Transposed matrix
+        R = np.stack((x, y, z), axis=-2)
+        t = - np.einsum('...ij, ...j', R, eye)
+    return R, t
+
+def look_at_for_habitat(eye, center, up, return_cam2world=True):
+    R, t = look_at(eye, center, up)
+    orientation = quaternion.from_rotation_matrix(R @ R_OPENCV2HABITAT.T)
+    return orientation, t
+
+def generate_orientation_noise(pan_range, tilt_range, roll_range):
+    return (quaternion.from_rotation_vector(np.random.uniform(*pan_range) * DEG2RAD * habitat_sim.geo.UP)
+            * quaternion.from_rotation_vector(np.random.uniform(*tilt_range) * DEG2RAD * habitat_sim.geo.RIGHT)
+            * quaternion.from_rotation_vector(np.random.uniform(*roll_range) * DEG2RAD * habitat_sim.geo.FRONT))
+
+
+class NoNaviguableSpaceError(RuntimeError):
+    def __init__(self, *args):
+            super().__init__(*args)
+
+class MultiviewHabitatSimGenerator:
+    def __init__(self,
+                scene,
+                navmesh,
+                scene_dataset_config_file,
+                resolution = (240, 320),
+                views_count=2,
+                hfov = 60,
+                gpu_id = 0,
+                size = 10000,
+                minimum_covisibility = 0.5,
+                transform = None):
+        self.scene = scene
+        self.navmesh = navmesh
+        self.scene_dataset_config_file = scene_dataset_config_file
+        self.resolution = resolution
+        self.views_count = views_count
+        assert(self.views_count >= 1)
+        self.hfov = hfov
+        self.gpu_id = gpu_id
+        self.size = size
+        self.transform = transform
+
+        # Noise added to camera orientation
+        self.pan_range = (-3, 3)
+        self.tilt_range = (-10, 10)
+        self.roll_range = (-5, 5)
+
+        # Height range to sample cameras
+        self.height_range = (1.2, 1.8)
+
+        # Random steps between the camera views
+        self.random_steps_count = 5
+        self.random_step_variance = 2.0
+
+        # Minimum fraction of the scene which should be valid (well defined depth)
+        self.minimum_valid_fraction = 0.7
+
+        # Distance threshold to see  to select pairs
+        self.distance_threshold = 0.05
+        # Minimum IoU of a view point cloud with respect to the reference view to be kept.
+        self.minimum_covisibility = minimum_covisibility
+
+        # Maximum number of retries.
+        self.max_attempts_count = 100
+
+        self.seed = None
+        self._lazy_initialization()
+
+    def _lazy_initialization(self):
+        # Lazy random seeding and instantiation of the simulator to deal with multiprocessing properly
+        if self.seed == None:
+            # Re-seed numpy generator
+            np.random.seed()
+            self.seed = np.random.randint(2**32-1)
+            sim_cfg = habitat_sim.SimulatorConfiguration()
+            sim_cfg.scene_id = self.scene
+            if self.scene_dataset_config_file is not None and self.scene_dataset_config_file != "":
+                    sim_cfg.scene_dataset_config_file = self.scene_dataset_config_file
+            sim_cfg.random_seed = self.seed
+            sim_cfg.load_semantic_mesh = False
+            sim_cfg.gpu_device_id = self.gpu_id
+
+            depth_sensor_spec = habitat_sim.CameraSensorSpec()
+            depth_sensor_spec.uuid = "depth"
+            depth_sensor_spec.sensor_type = habitat_sim.SensorType.DEPTH
+            depth_sensor_spec.resolution = self.resolution
+            depth_sensor_spec.hfov = self.hfov
+            depth_sensor_spec.position = [0.0, 0.0, 0]
+            depth_sensor_spec.orientation
+
+            rgb_sensor_spec = habitat_sim.CameraSensorSpec()
+            rgb_sensor_spec.uuid = "color"
+            rgb_sensor_spec.sensor_type = habitat_sim.SensorType.COLOR
+            rgb_sensor_spec.resolution = self.resolution
+            rgb_sensor_spec.hfov = self.hfov
+            rgb_sensor_spec.position = [0.0, 0.0, 0]
+            agent_cfg = habitat_sim.agent.AgentConfiguration(sensor_specifications=[rgb_sensor_spec, depth_sensor_spec])
+
+            cfg = habitat_sim.Configuration(sim_cfg, [agent_cfg])
+            self.sim = habitat_sim.Simulator(cfg)
+            if self.navmesh is not None and self.navmesh != "":
+                # Use pre-computed navmesh when available (usually better than those generated automatically)
+                self.sim.pathfinder.load_nav_mesh(self.navmesh)
+
+            if not self.sim.pathfinder.is_loaded:
+                # Try to compute a navmesh
+                navmesh_settings = habitat_sim.NavMeshSettings()
+                navmesh_settings.set_defaults()
+                self.sim.recompute_navmesh(self.sim.pathfinder, navmesh_settings, True)
+
+            # Ensure that the navmesh is not empty
+            if not self.sim.pathfinder.is_loaded:
+                raise NoNaviguableSpaceError(f"No naviguable location (scene: {self.scene} -- navmesh: {self.navmesh})")
+
+            self.agent = self.sim.initialize_agent(agent_id=0)
+
+    def close(self):
+        self.sim.close()
+
+    def __del__(self):
+        self.sim.close()
+
+    def __len__(self):
+        return self.size
+
+    def sample_random_viewpoint(self):
+        """ Sample a random viewpoint using the navmesh """
+        nav_point = self.sim.pathfinder.get_random_navigable_point()
+
+        # Sample a random viewpoint height
+        viewpoint_height = np.random.uniform(*self.height_range)
+        viewpoint_position = nav_point + viewpoint_height * habitat_sim.geo.UP
+        viewpoint_orientation = quaternion.from_rotation_vector(np.random.uniform(0, 2 * np.pi) * habitat_sim.geo.UP) * generate_orientation_noise(self.pan_range, self.tilt_range, self.roll_range)
+        return viewpoint_position, viewpoint_orientation, nav_point
+
+    def sample_other_random_viewpoint(self, observed_point, nav_point):
+        """ Sample a random viewpoint close to an existing one, using the navmesh and a reference observed point."""
+        other_nav_point = nav_point
+
+        walk_directions = self.random_step_variance * np.asarray([1,0,1])
+        for i in range(self.random_steps_count):
+            temp = self.sim.pathfinder.snap_point(other_nav_point + walk_directions * np.random.normal(size=3))
+            # Snapping may return nan when it fails
+            if not np.isnan(temp[0]):
+                    other_nav_point = temp
+
+        other_viewpoint_height = np.random.uniform(*self.height_range)
+        other_viewpoint_position = other_nav_point + other_viewpoint_height * habitat_sim.geo.UP
+
+        # Set viewing direction towards the central point
+        rotation, position = look_at_for_habitat(eye=other_viewpoint_position, center=observed_point, up=habitat_sim.geo.UP, return_cam2world=True)
+        rotation = rotation * generate_orientation_noise(self.pan_range, self.tilt_range, self.roll_range)
+        return position, rotation, other_nav_point
+
+    def is_other_pointcloud_overlapping(self, ref_pointcloud, other_pointcloud):
+        """ Check if a viewpoint is valid and overlaps significantly with a reference one. """
+        # Observation
+        pixels_count = self.resolution[0] * self.resolution[1]
+        valid_fraction = len(other_pointcloud) / pixels_count
+        assert valid_fraction <= 1.0 and valid_fraction >= 0.0
+        overlap = compute_pointcloud_overlaps_scikit(ref_pointcloud, other_pointcloud, self.distance_threshold, compute_symmetric=True)
+        covisibility = min(overlap["intersection1"] / pixels_count, overlap["intersection2"] / pixels_count)
+        is_valid = (valid_fraction >= self.minimum_valid_fraction) and (covisibility >= self.minimum_covisibility)
+        return is_valid, valid_fraction, covisibility
+
+    def is_other_viewpoint_overlapping(self, ref_pointcloud, observation, position, rotation):
+        """ Check if a viewpoint is valid and overlaps significantly with a reference one. """
+        # Observation
+        other_pointcloud = compute_pointcloud(observation['depth'], self.hfov, position, rotation)
+        return self.is_other_pointcloud_overlapping(ref_pointcloud, other_pointcloud)
+
+    def render_viewpoint(self, viewpoint_position, viewpoint_orientation):
+        agent_state = habitat_sim.AgentState()
+        agent_state.position = viewpoint_position
+        agent_state.rotation = viewpoint_orientation
+        self.agent.set_state(agent_state)
+        viewpoint_observations = self.sim.get_sensor_observations(agent_ids=0)
+        _append_camera_parameters(viewpoint_observations, self.hfov, viewpoint_position, viewpoint_orientation)
+        return viewpoint_observations
+
+    def __getitem__(self, useless_idx):
+        ref_position, ref_orientation, nav_point = self.sample_random_viewpoint()
+        ref_observations = self.render_viewpoint(ref_position, ref_orientation)
+        # Extract point cloud
+        ref_pointcloud = compute_pointcloud(depthmap=ref_observations['depth'], hfov=self.hfov,
+                                        camera_position=ref_position, camera_rotation=ref_orientation)
+
+        pixels_count = self.resolution[0] * self.resolution[1]
+        ref_valid_fraction = len(ref_pointcloud) / pixels_count
+        assert ref_valid_fraction <= 1.0 and ref_valid_fraction >= 0.0
+        if ref_valid_fraction < self.minimum_valid_fraction:
+                # This should produce a recursion error at some point when something is very wrong.
+                return self[0]
+        # Pick an reference observed point in the point cloud
+        observed_point = np.mean(ref_pointcloud, axis=0)
+
+        # Add the first image as reference
+        viewpoints_observations = [ref_observations]
+        viewpoints_covisibility = [ref_valid_fraction]
+        viewpoints_positions = [ref_position]
+        viewpoints_orientations = [quaternion.as_float_array(ref_orientation)]
+        viewpoints_clouds = [ref_pointcloud]
+        viewpoints_valid_fractions = [ref_valid_fraction]
+
+        for _ in range(self.views_count - 1):
+            # Generate an other viewpoint using some dummy random walk
+            successful_sampling = False
+            for sampling_attempt in range(self.max_attempts_count):
+                position, rotation, _ = self.sample_other_random_viewpoint(observed_point, nav_point)
+                # Observation
+                other_viewpoint_observations = self.render_viewpoint(position, rotation)
+                other_pointcloud = compute_pointcloud(other_viewpoint_observations['depth'], self.hfov, position, rotation)
+
+                is_valid, valid_fraction, covisibility = self.is_other_pointcloud_overlapping(ref_pointcloud, other_pointcloud)
+                if is_valid:
+                        successful_sampling = True
+                        break
+            if not successful_sampling:
+                print("WARNING: Maximum number of attempts reached.")
+                # Dirty hack, try using a novel original viewpoint
+                return self[0]
+            viewpoints_observations.append(other_viewpoint_observations)
+            viewpoints_covisibility.append(covisibility)
+            viewpoints_positions.append(position)
+            viewpoints_orientations.append(quaternion.as_float_array(rotation)) # WXYZ convention for the quaternion encoding.
+            viewpoints_clouds.append(other_pointcloud)
+            viewpoints_valid_fractions.append(valid_fraction)
+
+        # Estimate relations between all pairs of images
+        pairwise_visibility_ratios = np.ones((len(viewpoints_observations), len(viewpoints_observations)))
+        for i in range(len(viewpoints_observations)):
+            pairwise_visibility_ratios[i,i] = viewpoints_valid_fractions[i]
+            for j in range(i+1, len(viewpoints_observations)):
+                overlap = compute_pointcloud_overlaps_scikit(viewpoints_clouds[i], viewpoints_clouds[j], self.distance_threshold, compute_symmetric=True)
+                pairwise_visibility_ratios[i,j] = overlap['intersection1'] / pixels_count
+                pairwise_visibility_ratios[j,i] = overlap['intersection2'] / pixels_count
+
+        # IoU is relative to the image 0
+        data = {"observations": viewpoints_observations,
+                "positions": np.asarray(viewpoints_positions),
+                "orientations": np.asarray(viewpoints_orientations),
+                "covisibility_ratios": np.asarray(viewpoints_covisibility),
+                "valid_fractions": np.asarray(viewpoints_valid_fractions, dtype=float),
+                "pairwise_visibility_ratios": np.asarray(pairwise_visibility_ratios, dtype=float),
+                }
+
+        if self.transform is not None:
+            data = self.transform(data)
+        return  data
+
+    def generate_random_spiral_trajectory(self, images_count = 100, max_radius=0.5, half_turns=5, use_constant_orientation=False):
+        """
+        Return a list of images corresponding to a spiral trajectory from a random starting point.
+        Useful to generate nice visualisations.
+        Use an even number of half turns to get a nice "C1-continuous" loop effect 
+        """
+        ref_position, ref_orientation, navpoint = self.sample_random_viewpoint()
+        ref_observations = self.render_viewpoint(ref_position, ref_orientation)
+        ref_pointcloud = compute_pointcloud(depthmap=ref_observations['depth'], hfov=self.hfov,
+                                                        camera_position=ref_position, camera_rotation=ref_orientation)
+        pixels_count = self.resolution[0] * self.resolution[1]
+        if len(ref_pointcloud) / pixels_count < self.minimum_valid_fraction:
+            # Dirty hack: ensure that the valid part of the image is significant
+            return self.generate_random_spiral_trajectory(images_count, max_radius, half_turns, use_constant_orientation)
+
+        # Pick an observed point in the point cloud
+        observed_point = np.mean(ref_pointcloud, axis=0)
+        ref_R, ref_t = compute_camera_pose_opencv_convention(ref_position, ref_orientation)
+
+        images = []
+        is_valid = []
+        # Spiral trajectory, use_constant orientation
+        for i, alpha in enumerate(np.linspace(0, 1, images_count)):
+            r = max_radius * np.abs(np.sin(alpha * np.pi)) # Increase then decrease the radius
+            theta = alpha * half_turns * np.pi 
+            x = r * np.cos(theta)
+            y = r * np.sin(theta)
+            z = 0.0
+            position = ref_position + (ref_R @ np.asarray([x, y, z]).reshape(3,1)).flatten()
+            if use_constant_orientation:
+                orientation = ref_orientation
+            else:
+                # trajectory looking at a mean point in front of the ref observation
+                orientation, position = look_at_for_habitat(eye=position, center=observed_point, up=habitat_sim.geo.UP)
+            observations = self.render_viewpoint(position, orientation)
+            images.append(observations['color'][...,:3])
+            _is_valid, valid_fraction, iou = self.is_other_viewpoint_overlapping(ref_pointcloud, observations, position, orientation)
+            is_valid.append(_is_valid)
+        return images, np.all(is_valid)

croco/datasets/habitat_sim/pack_metadata_files.py

@@ -0,0 +1,69 @@
+# Copyright (C) 2022-present Naver Corporation. All rights reserved.
+# Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+"""
+Utility script to pack metadata files of the dataset in order to be able to re-generate it elsewhere.
+"""
+import os
+import glob
+from tqdm import tqdm
+import shutil
+import json
+from datasets.habitat_sim.paths import *
+import argparse
+import collections
+
+if __name__ == "__main__":
+    parser = argparse.ArgumentParser()
+    parser.add_argument("input_dir")
+    parser.add_argument("output_dir")
+    args = parser.parse_args()
+
+    input_dirname = args.input_dir
+    output_dirname = args.output_dir
+
+    input_metadata_filenames = glob.iglob(f"{input_dirname}/**/metadata.json", recursive=True)
+
+    images_count = collections.defaultdict(lambda : 0)
+    
+    os.makedirs(output_dirname)
+    for input_filename in tqdm(input_metadata_filenames):
+        # Ignore empty files
+        with open(input_filename, "r") as f:
+            original_metadata = json.load(f)
+            if "multiviews" not in original_metadata or len(original_metadata["multiviews"]) == 0:
+                print("No views in", input_filename)
+                continue
+
+        relpath = os.path.relpath(input_filename, input_dirname)
+        print(relpath)
+
+        # Copy metadata, while replacing scene paths by generic keys depending on the dataset, for portability.
+        # Data paths are sorted by decreasing length to avoid potential bugs due to paths starting by the same string pattern.
+        scenes_dataset_paths = dict(sorted(SCENES_DATASET.items(), key=lambda x: len(x[1]), reverse=True))
+        metadata = dict()
+        for key, value in original_metadata.items():
+            if key in ("scene_dataset_config_file", "scene", "navmesh") and value != "":
+                known_path = False
+                for dataset, dataset_path in scenes_dataset_paths.items():
+                    if value.startswith(dataset_path):
+                        value = os.path.join(dataset, os.path.relpath(value, dataset_path))
+                        known_path = True
+                        break
+                if not known_path:
+                    raise KeyError("Unknown path:" + value)
+            metadata[key] = value
+
+        # Compile some general statistics while packing data
+        scene_split = metadata["scene"].split("/")
+        upper_level = "/".join(scene_split[:2]) if scene_split[0] == "hm3d" else scene_split[0]
+        images_count[upper_level] += len(metadata["multiviews"])
+        
+        output_filename = os.path.join(output_dirname, relpath)
+        os.makedirs(os.path.dirname(output_filename), exist_ok=True)
+        with open(output_filename, "w") as f:
+            json.dump(metadata, f)
+
+    # Print statistics
+    print("Images count:")
+    for upper_level, count in images_count.items():
+        print(f"- {upper_level}: {count}")

croco/datasets/habitat_sim/paths.py

@@ -0,0 +1,129 @@
+# Copyright (C) 2022-present Naver Corporation. All rights reserved.
+# Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+
+"""
+Paths to Habitat-Sim scenes
+"""
+
+import os
+import json
+import collections
+from tqdm import tqdm
+
+
+# Hardcoded path to the different scene datasets
+SCENES_DATASET = {
+    "hm3d": "./data/habitat-sim-data/scene_datasets/hm3d/",
+    "gibson": "./data/habitat-sim-data/scene_datasets/gibson/",
+    "habitat-test-scenes": "./data/habitat-sim/scene_datasets/habitat-test-scenes/",
+    "replica_cad_baked_lighting": "./data/habitat-sim/scene_datasets/replica_cad_baked_lighting/",
+    "replica_cad": "./data/habitat-sim/scene_datasets/replica_cad/",
+    "replica": "./data/habitat-sim/scene_datasets/ReplicaDataset/",
+    "scannet": "./data/habitat-sim/scene_datasets/scannet/"
+}
+
+SceneData = collections.namedtuple("SceneData", ["scene_dataset_config_file", "scene", "navmesh", "output_dir"])
+
+def list_replicacad_scenes(base_output_dir, base_path=SCENES_DATASET["replica_cad"]):
+    scene_dataset_config_file = os.path.join(base_path, "replicaCAD.scene_dataset_config.json")
+    scenes = [f"apt_{i}" for i in range(6)] + ["empty_stage"]
+    navmeshes = [f"navmeshes/apt_{i}_static_furniture.navmesh" for i in range(6)] + ["empty_stage.navmesh"]
+    scenes_data = []
+    for idx in range(len(scenes)):
+        output_dir = os.path.join(base_output_dir, "ReplicaCAD", scenes[idx])
+        # Add scene
+        data = SceneData(scene_dataset_config_file=scene_dataset_config_file,
+                    scene = scenes[idx] + ".scene_instance.json",
+                    navmesh = os.path.join(base_path, navmeshes[idx]),
+                    output_dir = output_dir)
+        scenes_data.append(data)
+    return scenes_data
+
+def list_replica_cad_baked_lighting_scenes(base_output_dir, base_path=SCENES_DATASET["replica_cad_baked_lighting"]):
+    scene_dataset_config_file = os.path.join(base_path, "replicaCAD_baked.scene_dataset_config.json")
+    scenes = sum([[f"Baked_sc{i}_staging_{j:02}" for i in range(5)] for j in range(21)], [])
+    navmeshes = ""#[f"navmeshes/apt_{i}_static_furniture.navmesh" for i in range(6)] + ["empty_stage.navmesh"]
+    scenes_data = []
+    for idx in range(len(scenes)):
+        output_dir = os.path.join(base_output_dir, "replica_cad_baked_lighting", scenes[idx])
+        data = SceneData(scene_dataset_config_file=scene_dataset_config_file,
+                    scene = scenes[idx],
+                    navmesh = "",
+                    output_dir = output_dir)
+        scenes_data.append(data)
+    return scenes_data    
+
+def list_replica_scenes(base_output_dir, base_path):
+    scenes_data = []
+    for scene_id in os.listdir(base_path):
+        scene = os.path.join(base_path, scene_id, "mesh.ply")
+        navmesh = os.path.join(base_path, scene_id, "habitat/mesh_preseg_semantic.navmesh") # Not sure if I should use it
+        scene_dataset_config_file = ""
+        output_dir = os.path.join(base_output_dir, scene_id)
+        # Add scene only if it does not exist already, or if exist_ok
+        data = SceneData(scene_dataset_config_file = scene_dataset_config_file,
+                    scene = scene,
+                    navmesh = navmesh,
+                    output_dir = output_dir)
+        scenes_data.append(data)
+    return scenes_data
+
+
+def list_scenes(base_output_dir, base_path):
+    """
+    Generic method iterating through a base_path folder to find scenes.
+    """
+    scenes_data = []
+    for root, dirs, files in os.walk(base_path, followlinks=True):
+        folder_scenes_data = []
+        for file in files:
+            name, ext = os.path.splitext(file)
+            if ext == ".glb":
+                scene = os.path.join(root, name + ".glb")
+                navmesh = os.path.join(root, name + ".navmesh")
+                if not os.path.exists(navmesh):
+                    navmesh = ""
+                relpath = os.path.relpath(root, base_path)
+                output_dir = os.path.abspath(os.path.join(base_output_dir, relpath, name))
+                data = SceneData(scene_dataset_config_file="",
+                    scene = scene,
+                    navmesh = navmesh,
+                    output_dir = output_dir)
+                folder_scenes_data.append(data)
+
+        # Specific check for HM3D:
+        # When two meshesxxxx.basis.glb and xxxx.glb are present, use the 'basis' version.
+        basis_scenes = [data.scene[:-len(".basis.glb")] for data in folder_scenes_data if data.scene.endswith(".basis.glb")]
+        if len(basis_scenes) != 0:
+            folder_scenes_data = [data for data in folder_scenes_data if not (data.scene[:-len(".glb")] in basis_scenes)]
+
+        scenes_data.extend(folder_scenes_data)
+    return scenes_data
+
+def list_scenes_available(base_output_dir, scenes_dataset_paths=SCENES_DATASET):
+    scenes_data = []
+
+    # HM3D
+    for split in ("minival", "train", "val", "examples"):
+        scenes_data += list_scenes(base_output_dir=os.path.join(base_output_dir, f"hm3d/{split}/"),
+                                    base_path=f"{scenes_dataset_paths['hm3d']}/{split}")
+
+    # Gibson
+    scenes_data += list_scenes(base_output_dir=os.path.join(base_output_dir, "gibson"),
+                                base_path=scenes_dataset_paths["gibson"])
+
+    # Habitat test scenes (just a few)
+    scenes_data += list_scenes(base_output_dir=os.path.join(base_output_dir, "habitat-test-scenes"),
+                                base_path=scenes_dataset_paths["habitat-test-scenes"])
+
+    # ReplicaCAD (baked lightning)
+    scenes_data += list_replica_cad_baked_lighting_scenes(base_output_dir=base_output_dir)
+
+    # ScanNet
+    scenes_data += list_scenes(base_output_dir=os.path.join(base_output_dir, "scannet"), 
+                            base_path=scenes_dataset_paths["scannet"])
+    
+    # Replica
+    list_replica_scenes(base_output_dir=os.path.join(base_output_dir, "replica"),
+                        base_path=scenes_dataset_paths["replica"])
+    return scenes_data    

croco/datasets/pairs_dataset.py

@@ -0,0 +1,109 @@
+# Copyright (C) 2022-present Naver Corporation. All rights reserved.
+# Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+
+import os
+from torch.utils.data import Dataset
+from PIL import Image
+
+from datasets.transforms import get_pair_transforms
+
+def load_image(impath):
+    return Image.open(impath)
+
+def load_pairs_from_cache_file(fname, root=''):
+    assert os.path.isfile(fname), "cannot parse pairs from {:s}, file does not exist".format(fname)
+    with open(fname, 'r') as fid:
+        lines = fid.read().strip().splitlines()
+    pairs = [ (os.path.join(root,l.split()[0]), os.path.join(root,l.split()[1])) for l in lines]
+    return pairs
+    
+def load_pairs_from_list_file(fname, root=''):
+    assert os.path.isfile(fname), "cannot parse pairs from {:s}, file does not exist".format(fname)
+    with open(fname, 'r') as fid:
+        lines = fid.read().strip().splitlines()
+    pairs = [ (os.path.join(root,l+'_1.jpg'), os.path.join(root,l+'_2.jpg')) for l in lines if not l.startswith('#')]
+    return pairs
+    
+    
+def write_cache_file(fname, pairs, root=''):
+    if len(root)>0:
+        if not root.endswith('/'): root+='/'
+        assert os.path.isdir(root)
+    s = ''
+    for im1, im2 in pairs:
+        if len(root)>0:
+            assert im1.startswith(root), im1
+            assert im2.startswith(root), im2
+        s += '{:s} {:s}\n'.format(im1[len(root):], im2[len(root):])
+    with open(fname, 'w') as fid:
+        fid.write(s[:-1])
+    
+def parse_and_cache_all_pairs(dname, data_dir='./data/'):
+    if dname=='habitat_release':
+        dirname = os.path.join(data_dir, 'habitat_release')
+        assert os.path.isdir(dirname), "cannot find folder for habitat_release pairs: "+dirname
+        cache_file = os.path.join(dirname, 'pairs.txt')
+        assert not os.path.isfile(cache_file), "cache file already exists: "+cache_file
+        
+        print('Parsing pairs for dataset: '+dname)
+        pairs = []
+        for root, dirs, files in os.walk(dirname):
+            if 'val' in root: continue
+            dirs.sort()
+            pairs += [ (os.path.join(root,f), os.path.join(root,f[:-len('_1.jpeg')]+'_2.jpeg')) for f in sorted(files) if f.endswith('_1.jpeg')]
+        print('Found {:,} pairs'.format(len(pairs)))
+        print('Writing cache to: '+cache_file)
+        write_cache_file(cache_file, pairs, root=dirname)
+
+    else:
+        raise NotImplementedError('Unknown dataset: '+dname)
+    
+def dnames_to_image_pairs(dnames, data_dir='./data/'):
+    """
+    dnames: list of datasets with image pairs, separated by +
+    """
+    all_pairs = []
+    for dname in dnames.split('+'):
+        if dname=='habitat_release':
+            dirname = os.path.join(data_dir, 'habitat_release')
+            assert os.path.isdir(dirname), "cannot find folder for habitat_release pairs: "+dirname
+            cache_file = os.path.join(dirname, 'pairs.txt')
+            assert os.path.isfile(cache_file), "cannot find cache file for habitat_release pairs, please first create the cache file, see instructions. "+cache_file
+            pairs = load_pairs_from_cache_file(cache_file, root=dirname)
+        elif dname in ['ARKitScenes', 'MegaDepth', '3DStreetView', 'IndoorVL']:
+            dirname = os.path.join(data_dir, dname+'_crops')
+            assert os.path.isdir(dirname), "cannot find folder for {:s} pairs: {:s}".format(dname, dirname)
+            list_file = os.path.join(dirname, 'listing.txt')
+            assert os.path.isfile(list_file), "cannot find list file for {:s} pairs, see instructions. {:s}".format(dname, list_file)
+            pairs = load_pairs_from_list_file(list_file, root=dirname)            
+        print('  {:s}: {:,} pairs'.format(dname, len(pairs)))
+        all_pairs += pairs 
+    if '+' in dnames: print(' Total: {:,} pairs'.format(len(all_pairs)))
+    return all_pairs 
+
+
+class PairsDataset(Dataset):
+
+    def __init__(self, dnames, trfs='', totensor=True, normalize=True, data_dir='./data/'):
+        super().__init__()
+        self.image_pairs = dnames_to_image_pairs(dnames, data_dir=data_dir)
+        self.transforms = get_pair_transforms(transform_str=trfs, totensor=totensor, normalize=normalize)
+              
+    def __len__(self):
+        return len(self.image_pairs)
+            
+    def __getitem__(self, index):
+        im1path, im2path = self.image_pairs[index]
+        im1 = load_image(im1path)
+        im2 = load_image(im2path)
+        if self.transforms is not None: im1, im2 = self.transforms(im1, im2)
+        return im1, im2
+
+        
+if __name__=="__main__":
+    import argparse
+    parser = argparse.ArgumentParser(prog="Computing and caching list of pairs for a given dataset")
+    parser.add_argument('--data_dir', default='./data/', type=str, help="path where data are stored")
+    parser.add_argument('--dataset', default='habitat_release', type=str, help="name of the dataset")
+    args = parser.parse_args()
+    parse_and_cache_all_pairs(dname=args.dataset, data_dir=args.data_dir)

croco/datasets/transforms.py

@@ -0,0 +1,95 @@
+# Copyright (C) 2022-present Naver Corporation. All rights reserved.
+# Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+
+import torch
+import torchvision.transforms
+import torchvision.transforms.functional as F
+
+# "Pair": apply a transform on a pair
+# "Both": apply the exact same transform to both images
+
+class ComposePair(torchvision.transforms.Compose):
+    def __call__(self, img1, img2):
+        for t in self.transforms:
+            img1, img2 = t(img1, img2)
+        return img1, img2
+
+class NormalizeBoth(torchvision.transforms.Normalize):
+    def forward(self, img1, img2):
+        img1 = super().forward(img1)
+        img2 = super().forward(img2)
+        return img1, img2
+
+class ToTensorBoth(torchvision.transforms.ToTensor):
+    def __call__(self, img1, img2):
+        img1 = super().__call__(img1)
+        img2 = super().__call__(img2)
+        return img1, img2
+        
+class RandomCropPair(torchvision.transforms.RandomCrop): 
+    # the crop will be intentionally different for the two images with this class
+    def forward(self, img1, img2):
+        img1 = super().forward(img1)
+        img2 = super().forward(img2)
+        return img1, img2
+
+class ColorJitterPair(torchvision.transforms.ColorJitter): 
+    # can be symmetric (same for both images) or assymetric (different jitter params for each image) depending on assymetric_prob  
+    def __init__(self, assymetric_prob, **kwargs):
+        super().__init__(**kwargs)
+        self.assymetric_prob = assymetric_prob
+    def jitter_one(self, img, fn_idx, brightness_factor, contrast_factor, saturation_factor, hue_factor):
+        for fn_id in fn_idx:
+            if fn_id == 0 and brightness_factor is not None:
+                img = F.adjust_brightness(img, brightness_factor)
+            elif fn_id == 1 and contrast_factor is not None:
+                img = F.adjust_contrast(img, contrast_factor)
+            elif fn_id == 2 and saturation_factor is not None:
+                img = F.adjust_saturation(img, saturation_factor)
+            elif fn_id == 3 and hue_factor is not None:
+                img = F.adjust_hue(img, hue_factor)
+        return img
+        
+    def forward(self, img1, img2):
+
+        fn_idx, brightness_factor, contrast_factor, saturation_factor, hue_factor = self.get_params(
+            self.brightness, self.contrast, self.saturation, self.hue
+        )
+        img1 = self.jitter_one(img1, fn_idx, brightness_factor, contrast_factor, saturation_factor, hue_factor)
+        if torch.rand(1) < self.assymetric_prob: # assymetric:
+            fn_idx, brightness_factor, contrast_factor, saturation_factor, hue_factor = self.get_params(
+                self.brightness, self.contrast, self.saturation, self.hue
+            )
+        img2 = self.jitter_one(img2, fn_idx, brightness_factor, contrast_factor, saturation_factor, hue_factor)
+        return img1, img2
+
+def get_pair_transforms(transform_str, totensor=True, normalize=True):
+    # transform_str is eg    crop224+color
+    trfs = []
+    for s in transform_str.split('+'):
+        if s.startswith('crop'):
+            size = int(s[len('crop'):])
+            trfs.append(RandomCropPair(size))
+        elif s=='acolor':
+            trfs.append(ColorJitterPair(assymetric_prob=1.0, brightness=(0.6, 1.4), contrast=(0.6, 1.4), saturation=(0.6, 1.4), hue=0.0))
+        elif s=='': # if transform_str was ""
+            pass
+        else:
+            raise NotImplementedError('Unknown augmentation: '+s)
+            
+    if totensor:
+        trfs.append( ToTensorBoth() )
+    if normalize:
+        trfs.append( NormalizeBoth(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]) )
+
+    if len(trfs)==0:
+        return None
+    elif len(trfs)==1:
+        return trfs
+    else:
+        return ComposePair(trfs)
+        
+        
+        
+        
+        

croco/demo.py

@@ -0,0 +1,55 @@
+# Copyright (C) 2022-present Naver Corporation. All rights reserved.
+# Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+
+import torch
+from models.croco import CroCoNet
+from PIL import Image
+import torchvision.transforms
+from torchvision.transforms import ToTensor, Normalize, Compose
+
+def main():
+    device = torch.device('cuda:0' if torch.cuda.is_available() and torch.cuda.device_count()>0 else 'cpu')
+    
+    # load 224x224 images and transform them to tensor 
+    imagenet_mean = [0.485, 0.456, 0.406]
+    imagenet_mean_tensor = torch.tensor(imagenet_mean).view(1,3,1,1).to(device, non_blocking=True)
+    imagenet_std = [0.229, 0.224, 0.225]
+    imagenet_std_tensor = torch.tensor(imagenet_std).view(1,3,1,1).to(device, non_blocking=True)
+    trfs = Compose([ToTensor(), Normalize(mean=imagenet_mean, std=imagenet_std)])
+    image1 = trfs(Image.open('assets/Chateau1.png').convert('RGB')).to(device, non_blocking=True).unsqueeze(0)
+    image2 = trfs(Image.open('assets/Chateau2.png').convert('RGB')).to(device, non_blocking=True).unsqueeze(0)
+    
+    # load model 
+    ckpt = torch.load('pretrained_models/CroCo_V2_ViTLarge_BaseDecoder.pth', 'cpu')
+    model = CroCoNet( **ckpt.get('croco_kwargs',{})).to(device)
+    model.eval()
+    msg = model.load_state_dict(ckpt['model'], strict=True)
+    
+    # forward 
+    with torch.inference_mode():
+        out, mask, target = model(image1, image2)
+        
+    # the output is normalized, thus use the mean/std of the actual image to go back to RGB space 
+    patchified = model.patchify(image1)
+    mean = patchified.mean(dim=-1, keepdim=True)
+    var = patchified.var(dim=-1, keepdim=True)
+    decoded_image = model.unpatchify(out * (var + 1.e-6)**.5 + mean)
+    # undo imagenet normalization, prepare masked image
+    decoded_image = decoded_image * imagenet_std_tensor + imagenet_mean_tensor
+    input_image = image1 * imagenet_std_tensor + imagenet_mean_tensor
+    ref_image = image2 * imagenet_std_tensor + imagenet_mean_tensor
+    image_masks = model.unpatchify(model.patchify(torch.ones_like(ref_image)) * mask[:,:,None])
+    masked_input_image = ((1 - image_masks) * input_image)
+
+    # make visualization
+    visualization = torch.cat((ref_image, masked_input_image, decoded_image, input_image), dim=3) # 4*(B, 3, H, W) -> B, 3, H, W*4
+    B, C, H, W = visualization.shape
+    visualization = visualization.permute(1, 0, 2, 3).reshape(C, B*H, W)
+    visualization = torchvision.transforms.functional.to_pil_image(torch.clamp(visualization, 0, 1))
+    fname = "demo_output.png"
+    visualization.save(fname)
+    print('Visualization save in '+fname)
+    
+
+if __name__=="__main__":
+    main()

croco/interactive_demo.ipynb

@@ -0,0 +1,271 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Interactive demo of Cross-view Completion."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "# Copyright (C) 2022-present Naver Corporation. All rights reserved.\n",
+    "# Licensed under CC BY-NC-SA 4.0 (non-commercial use only)."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import torch\n",
+    "import numpy as np\n",
+    "from models.croco import CroCoNet\n",
+    "from ipywidgets import interact, interactive, fixed, interact_manual\n",
+    "import ipywidgets as widgets\n",
+    "import matplotlib.pyplot as plt\n",
+    "import quaternion\n",
+    "import models.masking"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Load CroCo model"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "ckpt = torch.load('pretrained_models/CroCo_V2_ViTLarge_BaseDecoder.pth', 'cpu')\n",
+    "model = CroCoNet( **ckpt.get('croco_kwargs',{}))\n",
+    "msg = model.load_state_dict(ckpt['model'], strict=True)\n",
+    "use_gpu = torch.cuda.is_available() and torch.cuda.device_count()>0\n",
+    "device = torch.device('cuda:0' if use_gpu else 'cpu')\n",
+    "model = model.eval()\n",
+    "model = model.to(device=device)\n",
+    "print(msg)\n",
+    "\n",
+    "def process_images(ref_image, target_image, masking_ratio, reconstruct_unmasked_patches=False):\n",
+    "    \"\"\"\n",
+    "    Perform Cross-View completion using two input images, specified using Numpy arrays.\n",
+    "    \"\"\"\n",
+    "    # Replace the mask generator\n",
+    "    model.mask_generator = models.masking.RandomMask(model.patch_embed.num_patches, masking_ratio)\n",
+    "\n",
+    "    # ImageNet-1k color normalization\n",
+    "    imagenet_mean = torch.as_tensor([0.485, 0.456, 0.406]).reshape(1,3,1,1).to(device)\n",
+    "    imagenet_std = torch.as_tensor([0.229, 0.224, 0.225]).reshape(1,3,1,1).to(device)\n",
+    "\n",
+    "    normalize_input_colors = True\n",
+    "    is_output_normalized = True\n",
+    "    with torch.no_grad():\n",
+    "        # Cast data to torch\n",
+    "        target_image = (torch.as_tensor(target_image, dtype=torch.float, device=device).permute(2,0,1) / 255)[None]\n",
+    "        ref_image = (torch.as_tensor(ref_image, dtype=torch.float, device=device).permute(2,0,1) / 255)[None]\n",
+    "\n",
+    "        if normalize_input_colors:\n",
+    "            ref_image = (ref_image - imagenet_mean) / imagenet_std\n",
+    "            target_image = (target_image - imagenet_mean) / imagenet_std\n",
+    "\n",
+    "        out, mask, _ = model(target_image, ref_image)\n",
+    "        # # get target\n",
+    "        if not is_output_normalized:\n",
+    "            predicted_image = model.unpatchify(out)\n",
+    "        else:\n",
+    "            # The output only contains higher order information,\n",
+    "            # we retrieve mean and standard deviation from the actual target image\n",
+    "            patchified = model.patchify(target_image)\n",
+    "            mean = patchified.mean(dim=-1, keepdim=True)\n",
+    "            var = patchified.var(dim=-1, keepdim=True)\n",
+    "            pred_renorm = out * (var + 1.e-6)**.5 + mean\n",
+    "            predicted_image = model.unpatchify(pred_renorm)\n",
+    "\n",
+    "        image_masks = model.unpatchify(model.patchify(torch.ones_like(ref_image)) * mask[:,:,None])\n",
+    "        masked_target_image = (1 - image_masks) * target_image\n",
+    "      \n",
+    "        if not reconstruct_unmasked_patches:\n",
+    "            # Replace unmasked patches by their actual values\n",
+    "            predicted_image = predicted_image * image_masks + masked_target_image\n",
+    "\n",
+    "        # Unapply color normalization\n",
+    "        if normalize_input_colors:\n",
+    "            predicted_image = predicted_image * imagenet_std + imagenet_mean\n",
+    "            masked_target_image = masked_target_image * imagenet_std + imagenet_mean\n",
+    "        \n",
+    "        # Cast to Numpy\n",
+    "        masked_target_image = np.asarray(torch.clamp(masked_target_image.squeeze(0).permute(1,2,0) * 255, 0, 255).cpu().numpy(), dtype=np.uint8)\n",
+    "        predicted_image = np.asarray(torch.clamp(predicted_image.squeeze(0).permute(1,2,0) * 255, 0, 255).cpu().numpy(), dtype=np.uint8)\n",
+    "        return masked_target_image, predicted_image"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Use the Habitat simulator to render images from arbitrary viewpoints (requires habitat_sim to be installed)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import os\n",
+    "os.environ[\"MAGNUM_LOG\"]=\"quiet\"\n",
+    "os.environ[\"HABITAT_SIM_LOG\"]=\"quiet\"\n",
+    "import habitat_sim\n",
+    "\n",
+    "scene = \"habitat-sim-data/scene_datasets/habitat-test-scenes/skokloster-castle.glb\"\n",
+    "navmesh = \"habitat-sim-data/scene_datasets/habitat-test-scenes/skokloster-castle.navmesh\"\n",
+    "\n",
+    "sim_cfg = habitat_sim.SimulatorConfiguration()\n",
+    "if use_gpu: sim_cfg.gpu_device_id = 0\n",
+    "sim_cfg.scene_id = scene\n",
+    "sim_cfg.load_semantic_mesh = False\n",
+    "rgb_sensor_spec = habitat_sim.CameraSensorSpec()\n",
+    "rgb_sensor_spec.uuid = \"color\"\n",
+    "rgb_sensor_spec.sensor_type = habitat_sim.SensorType.COLOR\n",
+    "rgb_sensor_spec.resolution = (224,224)\n",
+    "rgb_sensor_spec.hfov = 56.56\n",
+    "rgb_sensor_spec.position = [0.0, 0.0, 0.0]\n",
+    "rgb_sensor_spec.orientation = [0, 0, 0]\n",
+    "agent_cfg = habitat_sim.agent.AgentConfiguration(sensor_specifications=[rgb_sensor_spec])\n",
+    "\n",
+    "\n",
+    "cfg = habitat_sim.Configuration(sim_cfg, [agent_cfg])\n",
+    "sim = habitat_sim.Simulator(cfg)\n",
+    "if navmesh is not None:\n",
+    "    sim.pathfinder.load_nav_mesh(navmesh)\n",
+    "agent = sim.initialize_agent(agent_id=0)\n",
+    "\n",
+    "def sample_random_viewpoint():\n",
+    "    \"\"\" Sample a random viewpoint using the navmesh \"\"\"\n",
+    "    nav_point = sim.pathfinder.get_random_navigable_point()\n",
+    "    # Sample a random viewpoint height\n",
+    "    viewpoint_height = np.random.uniform(1.0, 1.6)\n",
+    "    viewpoint_position = nav_point + viewpoint_height * habitat_sim.geo.UP\n",
+    "    viewpoint_orientation = quaternion.from_rotation_vector(np.random.uniform(-np.pi, np.pi) * habitat_sim.geo.UP)\n",
+    "    return viewpoint_position, viewpoint_orientation\n",
+    "\n",
+    "def render_viewpoint(position, orientation):\n",
+    "    agent_state = habitat_sim.AgentState()\n",
+    "    agent_state.position = position\n",
+    "    agent_state.rotation = orientation\n",
+    "    agent.set_state(agent_state)\n",
+    "    viewpoint_observations = sim.get_sensor_observations(agent_ids=0)\n",
+    "    image = viewpoint_observations['color'][:,:,:3]\n",
+    "    image = np.asarray(np.clip(1.5 * np.asarray(image, dtype=float), 0, 255), dtype=np.uint8)\n",
+    "    return image"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Sample a random reference view"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "ref_position, ref_orientation = sample_random_viewpoint()\n",
+    "ref_image = render_viewpoint(ref_position, ref_orientation)\n",
+    "plt.clf()\n",
+    "fig, axes = plt.subplots(1,1, squeeze=False, num=1)\n",
+    "axes[0,0].imshow(ref_image)\n",
+    "for ax in axes.flatten():\n",
+    "    ax.set_xticks([])\n",
+    "    ax.set_yticks([])"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "### Interactive cross-view completion using CroCo"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "reconstruct_unmasked_patches = False\n",
+    "\n",
+    "def show_demo(masking_ratio, x, y, z, panorama, elevation):\n",
+    "    R = quaternion.as_rotation_matrix(ref_orientation)\n",
+    "    target_position = ref_position + x * R[:,0] + y * R[:,1] + z * R[:,2]\n",
+    "    target_orientation = (ref_orientation\n",
+    "         * quaternion.from_rotation_vector(-elevation * np.pi/180 * habitat_sim.geo.LEFT) \n",
+    "         * quaternion.from_rotation_vector(-panorama * np.pi/180 * habitat_sim.geo.UP))\n",
+    "    \n",
+    "    ref_image = render_viewpoint(ref_position, ref_orientation)\n",
+    "    target_image = render_viewpoint(target_position, target_orientation)\n",
+    "\n",
+    "    masked_target_image, predicted_image = process_images(ref_image, target_image, masking_ratio, reconstruct_unmasked_patches)\n",
+    "\n",
+    "    fig, axes = plt.subplots(1,4, squeeze=True, dpi=300)\n",
+    "    axes[0].imshow(ref_image)\n",
+    "    axes[0].set_xlabel(\"Reference\")\n",
+    "    axes[1].imshow(masked_target_image)\n",
+    "    axes[1].set_xlabel(\"Masked target\")\n",
+    "    axes[2].imshow(predicted_image)\n",
+    "    axes[2].set_xlabel(\"Reconstruction\")        \n",
+    "    axes[3].imshow(target_image)\n",
+    "    axes[3].set_xlabel(\"Target\")\n",
+    "    for ax in axes.flatten():\n",
+    "        ax.set_xticks([])\n",
+    "        ax.set_yticks([])\n",
+    "\n",
+    "interact(show_demo,\n",
+    "        masking_ratio=widgets.FloatSlider(description='masking', value=0.9, min=0.0, max=1.0),\n",
+    "        x=widgets.FloatSlider(value=0.0, min=-0.5, max=0.5, step=0.05),\n",
+    "        y=widgets.FloatSlider(value=0.0, min=-0.5, max=0.5, step=0.05),\n",
+    "        z=widgets.FloatSlider(value=0.0, min=-0.5, max=0.5, step=0.05),\n",
+    "        panorama=widgets.FloatSlider(value=0.0, min=-20, max=20, step=0.5),\n",
+    "        elevation=widgets.FloatSlider(value=0.0, min=-20, max=20, step=0.5));"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.7.13"
+  },
+  "vscode": {
+   "interpreter": {
+    "hash": "f9237820cd248d7e07cb4fb9f0e4508a85d642f19d831560c0a4b61f3e907e67"
+   }
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 2
+}

croco/models/blocks.py

@@ -0,0 +1,241 @@
+# Copyright (C) 2022-present Naver Corporation. All rights reserved.
+# Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+
+
+# --------------------------------------------------------
+# Main encoder/decoder blocks
+# --------------------------------------------------------
+# References: 
+# timm
+# https://github.com/rwightman/pytorch-image-models/blob/master/timm/models/vision_transformer.py
+# https://github.com/rwightman/pytorch-image-models/blob/master/timm/models/layers/helpers.py
+# https://github.com/rwightman/pytorch-image-models/blob/master/timm/models/layers/drop.py
+# https://github.com/rwightman/pytorch-image-models/blob/master/timm/models/layers/mlp.py
+# https://github.com/rwightman/pytorch-image-models/blob/master/timm/models/layers/patch_embed.py
+
+
+import torch
+import torch.nn as nn 
+
+from itertools import repeat
+import collections.abc
+
+
+def _ntuple(n):
+    def parse(x):
+        if isinstance(x, collections.abc.Iterable) and not isinstance(x, str):
+            return x
+        return tuple(repeat(x, n))
+    return parse
+to_2tuple = _ntuple(2)
+
+def drop_path(x, drop_prob: float = 0., training: bool = False, scale_by_keep: bool = True):
+    """Drop paths (Stochastic Depth) per sample (when applied in main path of residual blocks).
+    """
+    if drop_prob == 0. or not training:
+        return x
+    keep_prob = 1 - drop_prob
+    shape = (x.shape[0],) + (1,) * (x.ndim - 1)  # work with diff dim tensors, not just 2D ConvNets
+    random_tensor = x.new_empty(shape).bernoulli_(keep_prob)
+    if keep_prob > 0.0 and scale_by_keep:
+        random_tensor.div_(keep_prob)
+    return x * random_tensor
+
+class DropPath(nn.Module):
+    """Drop paths (Stochastic Depth) per sample  (when applied in main path of residual blocks).
+    """
+    def __init__(self, drop_prob: float = 0., scale_by_keep: bool = True):
+        super(DropPath, self).__init__()
+        self.drop_prob = drop_prob
+        self.scale_by_keep = scale_by_keep
+
+    def forward(self, x):
+        return drop_path(x, self.drop_prob, self.training, self.scale_by_keep)
+
+    def extra_repr(self):
+        return f'drop_prob={round(self.drop_prob,3):0.3f}'
+
+class Mlp(nn.Module):
+    """ MLP as used in Vision Transformer, MLP-Mixer and related networks"""
+    def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, bias=True, drop=0.):
+        super().__init__()
+        out_features = out_features or in_features
+        hidden_features = hidden_features or in_features
+        bias = to_2tuple(bias)
+        drop_probs = to_2tuple(drop)
+
+        self.fc1 = nn.Linear(in_features, hidden_features, bias=bias[0])
+        self.act = act_layer()
+        self.drop1 = nn.Dropout(drop_probs[0])
+        self.fc2 = nn.Linear(hidden_features, out_features, bias=bias[1])
+        self.drop2 = nn.Dropout(drop_probs[1])
+
+    def forward(self, x):
+        x = self.fc1(x)
+        x = self.act(x)
+        x = self.drop1(x)
+        x = self.fc2(x)
+        x = self.drop2(x)
+        return x
+
+class Attention(nn.Module):
+
+    def __init__(self, dim, rope=None, num_heads=8, qkv_bias=False, attn_drop=0., proj_drop=0.):
+        super().__init__()
+        self.num_heads = num_heads
+        head_dim = dim // num_heads
+        self.scale = head_dim ** -0.5
+        self.qkv = nn.Linear(dim, dim * 3, bias=qkv_bias)
+        self.attn_drop = nn.Dropout(attn_drop)
+        self.proj = nn.Linear(dim, dim)
+        self.proj_drop = nn.Dropout(proj_drop)
+        self.rope = rope 
+
+    def forward(self, x, xpos):
+        B, N, C = x.shape
+
+        qkv = self.qkv(x).reshape(B, N, 3, self.num_heads, C // self.num_heads).transpose(1,3)
+        q, k, v = [qkv[:,:,i] for i in range(3)]
+        # q,k,v = qkv.unbind(2)  # make torchscript happy (cannot use tensor as tuple)
+               
+        if self.rope is not None:
+            q = self.rope(q, xpos)
+            k = self.rope(k, xpos)
+               
+        attn = (q @ k.transpose(-2, -1)) * self.scale
+        attn = attn.softmax(dim=-1)
+        attn = self.attn_drop(attn)
+
+        x = (attn @ v).transpose(1, 2).reshape(B, N, C)
+        x = self.proj(x)
+        x = self.proj_drop(x)
+        return x
+
+class Block(nn.Module):
+
+    def __init__(self, dim, num_heads, mlp_ratio=4., qkv_bias=False, drop=0., attn_drop=0.,
+                 drop_path=0., act_layer=nn.GELU, norm_layer=nn.LayerNorm, rope=None):
+        super().__init__()
+        self.norm1 = norm_layer(dim)
+        self.attn = Attention(dim, rope=rope, num_heads=num_heads, qkv_bias=qkv_bias, attn_drop=attn_drop, proj_drop=drop)
+        # NOTE: drop path for stochastic depth, we shall see if this is better than dropout here
+        self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity()
+        self.norm2 = norm_layer(dim)
+        mlp_hidden_dim = int(dim * mlp_ratio)
+        self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim, act_layer=act_layer, drop=drop)
+
+    def forward(self, x, xpos):
+        x = x + self.drop_path(self.attn(self.norm1(x), xpos))
+        x = x + self.drop_path(self.mlp(self.norm2(x)))
+        return x
+
+class CrossAttention(nn.Module):
+    
+    def __init__(self, dim, rope=None, num_heads=8, qkv_bias=False, attn_drop=0., proj_drop=0.):
+        super().__init__()
+        self.num_heads = num_heads
+        head_dim = dim // num_heads
+        self.scale = head_dim ** -0.5
+
+        self.projq = nn.Linear(dim, dim, bias=qkv_bias)
+        self.projk = nn.Linear(dim, dim, bias=qkv_bias)
+        self.projv = nn.Linear(dim, dim, bias=qkv_bias)
+        self.attn_drop = nn.Dropout(attn_drop)
+        self.proj = nn.Linear(dim, dim)
+        self.proj_drop = nn.Dropout(proj_drop)
+        
+        self.rope = rope
+        
+    def forward(self, query, key, value, qpos, kpos):
+        B, Nq, C = query.shape
+        Nk = key.shape[1]
+        Nv = value.shape[1]
+        
+        q = self.projq(query).reshape(B,Nq,self.num_heads, C// self.num_heads).permute(0, 2, 1, 3)
+        k = self.projk(key).reshape(B,Nk,self.num_heads, C// self.num_heads).permute(0, 2, 1, 3)
+        v = self.projv(value).reshape(B,Nv,self.num_heads, C// self.num_heads).permute(0, 2, 1, 3)
+        
+        if self.rope is not None:
+            q = self.rope(q, qpos)
+            k = self.rope(k, kpos)
+            
+        attn = (q @ k.transpose(-2, -1)) * self.scale
+        attn = attn.softmax(dim=-1)
+        attn = self.attn_drop(attn)
+
+        x = (attn @ v).transpose(1, 2).reshape(B, Nq, C)
+        x = self.proj(x)
+        x = self.proj_drop(x)
+        return x
+
+class DecoderBlock(nn.Module):
+
+    def __init__(self, dim, num_heads, mlp_ratio=4., qkv_bias=False, drop=0., attn_drop=0.,
+                 drop_path=0., act_layer=nn.GELU, norm_layer=nn.LayerNorm, norm_mem=True, rope=None):
+        super().__init__()
+        self.norm1 = norm_layer(dim)
+        self.attn = Attention(dim, rope=rope, num_heads=num_heads, qkv_bias=qkv_bias, attn_drop=attn_drop, proj_drop=drop)
+        self.cross_attn = CrossAttention(dim, rope=rope, num_heads=num_heads, qkv_bias=qkv_bias, attn_drop=attn_drop, proj_drop=drop)
+        self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity()
+        self.norm2 = norm_layer(dim)
+        self.norm3 = norm_layer(dim)
+        mlp_hidden_dim = int(dim * mlp_ratio)
+        self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim, act_layer=act_layer, drop=drop)
+        self.norm_y = norm_layer(dim) if norm_mem else nn.Identity()
+
+    def forward(self, x, y, xpos, ypos):
+        x = x + self.drop_path(self.attn(self.norm1(x), xpos))
+        y_ = self.norm_y(y)
+        x = x + self.drop_path(self.cross_attn(self.norm2(x), y_, y_, xpos, ypos))
+        x = x + self.drop_path(self.mlp(self.norm3(x)))
+        return x, y
+        
+        
+# patch embedding
+class PositionGetter(object):
+    """ return positions of patches """
+
+    def __init__(self):
+        self.cache_positions = {}
+        
+    def __call__(self, b, h, w, device):
+        if not (h,w) in self.cache_positions:
+            x = torch.arange(w, device=device)
+            y = torch.arange(h, device=device)
+            self.cache_positions[h,w] = torch.cartesian_prod(y, x) # (h, w, 2)
+        pos = self.cache_positions[h,w].view(1, h*w, 2).expand(b, -1, 2).clone()
+        return pos
+
+class PatchEmbed(nn.Module):
+    """ just adding _init_weights + position getter compared to timm.models.layers.patch_embed.PatchEmbed"""
+
+    def __init__(self, img_size=224, patch_size=16, in_chans=3, embed_dim=768, norm_layer=None, flatten=True):
+        super().__init__()
+        img_size = to_2tuple(img_size)
+        patch_size = to_2tuple(patch_size)
+        self.img_size = img_size
+        self.patch_size = patch_size
+        self.grid_size = (img_size[0] // patch_size[0], img_size[1] // patch_size[1])
+        self.num_patches = self.grid_size[0] * self.grid_size[1]
+        self.flatten = flatten
+
+        self.proj = nn.Conv2d(in_chans, embed_dim, kernel_size=patch_size, stride=patch_size)
+        self.norm = norm_layer(embed_dim) if norm_layer else nn.Identity()
+        
+        self.position_getter = PositionGetter()
+        
+    def forward(self, x):
+        B, C, H, W = x.shape
+        torch._assert(H == self.img_size[0], f"Input image height ({H}) doesn't match model ({self.img_size[0]}).")
+        torch._assert(W == self.img_size[1], f"Input image width ({W}) doesn't match model ({self.img_size[1]}).")
+        x = self.proj(x)
+        pos = self.position_getter(B, x.size(2), x.size(3), x.device)
+        if self.flatten:
+            x = x.flatten(2).transpose(1, 2)  # BCHW -> BNC
+        x = self.norm(x)
+        return x, pos
+        
+    def _init_weights(self):
+        w = self.proj.weight.data
+        torch.nn.init.xavier_uniform_(w.view([w.shape[0], -1])) 
+

croco/models/criterion.py

@@ -0,0 +1,37 @@
+# Copyright (C) 2022-present Naver Corporation. All rights reserved.
+# Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+# 
+# --------------------------------------------------------
+# Criterion to train CroCo
+# --------------------------------------------------------
+# References:
+# MAE: https://github.com/facebookresearch/mae
+# --------------------------------------------------------
+
+import torch
+
+class MaskedMSE(torch.nn.Module):
+
+    def __init__(self, norm_pix_loss=False, masked=True):
+        """
+            norm_pix_loss: normalize each patch by their pixel mean and variance
+            masked: compute loss over the masked patches only 
+        """
+        super().__init__()
+        self.norm_pix_loss = norm_pix_loss
+        self.masked = masked 
+        
+    def forward(self, pred, mask, target):
+        
+        if self.norm_pix_loss:
+            mean = target.mean(dim=-1, keepdim=True)
+            var = target.var(dim=-1, keepdim=True)
+            target = (target - mean) / (var + 1.e-6)**.5
+            
+        loss = (pred - target) ** 2
+        loss = loss.mean(dim=-1)  # [N, L], mean loss per patch
+        if self.masked:
+            loss = (loss * mask).sum() / mask.sum()  # mean loss on masked patches
+        else:
+            loss = loss.mean()  # mean loss
+        return loss

croco/models/croco.py

@@ -0,0 +1,256 @@
+# Copyright (C) 2022-present Naver Corporation. All rights reserved.
+# Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+
+
+# --------------------------------------------------------
+# CroCo model during pretraining
+# --------------------------------------------------------
+
+
+
+import torch
+import torch.nn as nn
+torch.backends.cuda.matmul.allow_tf32 = True # for gpu >= Ampere and pytorch >= 1.12
+from functools import partial
+
+from models.blocks import Block, DecoderBlock, PatchEmbed
+from models.pos_embed import get_2d_sincos_pos_embed, RoPE2D 
+from models.masking import RandomMask
+
+
+class CroCoNet(nn.Module):
+
+    def __init__(self,
+                 img_size=224,           # input image size
+                 patch_size=16,          # patch_size 
+                 mask_ratio=0.9,         # ratios of masked tokens 
+                 enc_embed_dim=768,      # encoder feature dimension
+                 enc_depth=12,           # encoder depth 
+                 enc_num_heads=12,       # encoder number of heads in the transformer block 
+                 dec_embed_dim=512,      # decoder feature dimension 
+                 dec_depth=8,            # decoder depth 
+                 dec_num_heads=16,       # decoder number of heads in the transformer block 
+                 mlp_ratio=4,
+                 norm_layer=partial(nn.LayerNorm, eps=1e-6),
+                 norm_im2_in_dec=True,   # whether to apply normalization of the 'memory' = (second image) in the decoder 
+                 pos_embed='cosine',     # positional embedding (either cosine or RoPE100)
+                ):
+                
+        super(CroCoNet, self).__init__()
+
+        self.enc_depth = enc_depth
+        self.enc_embed_dim = enc_embed_dim
+        self.dec_depth = dec_depth
+        self.dec_embed_dim = dec_embed_dim
+        # patch embeddings  (with initialization done as in MAE)
+        self._set_patch_embed(img_size, patch_size, enc_embed_dim)
+
+        # mask generations
+        self._set_mask_generator(self.patch_embed.num_patches, mask_ratio)
+
+        self.pos_embed = pos_embed
+        if pos_embed=='cosine':
+            # positional embedding of the encoder 
+            enc_pos_embed = get_2d_sincos_pos_embed(enc_embed_dim, self.patch_embed.grid_size, n_cls_token=0)
+            self.register_buffer('enc_pos_embed', torch.from_numpy(enc_pos_embed).float())
+            # positional embedding of the decoder  
+            dec_pos_embed = get_2d_sincos_pos_embed(dec_embed_dim, self.patch_embed.grid_size, n_cls_token=0)
+            self.register_buffer('dec_pos_embed', torch.from_numpy(dec_pos_embed).float())
+            # pos embedding in each block
+            self.rope = None # nothing for cosine 
+        elif pos_embed.startswith('RoPE'): # eg RoPE100 
+            self.enc_pos_embed = None # nothing to add in the encoder with RoPE
+            self.dec_pos_embed = None # nothing to add in the decoder with RoPE
+            if RoPE2D is None: raise ImportError("Cannot find cuRoPE2D, please install it following the README instructions")
+            freq = float(pos_embed[len('RoPE'):])
+            self.rope = RoPE2D(freq=freq)
+        else:
+            raise NotImplementedError('Unknown pos_embed '+pos_embed)
+
+        # transformer for the encoder 
+
+        self.enc_blocks = nn.ModuleList([
+            Block(enc_embed_dim, enc_num_heads, mlp_ratio, qkv_bias=True, norm_layer=norm_layer, rope=self.rope)
+            for i in range(enc_depth)])
+        self.enc_norm = norm_layer(enc_embed_dim)
+
+
+        self.dec_blocks_pc = nn.ModuleList([
+            Block(dec_embed_dim, dec_num_heads, mlp_ratio, qkv_bias=True, norm_layer=norm_layer, rope=self.rope)
+            for i in range(dec_depth//2-2)])
+        # masked tokens 
+        self._set_mask_token(dec_embed_dim)
+
+        # decoder 
+        self._set_decoder(enc_embed_dim, dec_embed_dim, dec_num_heads, dec_depth, mlp_ratio, norm_layer, norm_im2_in_dec)
+        
+        # prediction head 
+        self._set_prediction_head(dec_embed_dim, patch_size)
+        
+        # initializer weights
+        self.initialize_weights()           
+
+    def _set_patch_embed(self, img_size=224, patch_size=16, enc_embed_dim=768):
+        self.patch_embed = PatchEmbed(img_size, patch_size, 3, enc_embed_dim)
+
+    def _set_mask_generator(self, num_patches, mask_ratio):
+        self.mask_generator = RandomMask(num_patches, mask_ratio)
+        
+    def _set_mask_token(self, dec_embed_dim):
+        self.mask_token = nn.Parameter(torch.zeros(1, 1, dec_embed_dim))
+        
+    def _set_decoder(self, enc_embed_dim, dec_embed_dim, dec_num_heads, dec_depth, mlp_ratio, norm_layer, norm_im2_in_dec):
+        self.dec_depth = dec_depth
+        self.dec_embed_dim = dec_embed_dim
+        # transfer from encoder to decoder 
+        self.decoder_embed = nn.Linear(enc_embed_dim, dec_embed_dim, bias=True)
+        # transformer for the decoder 
+        self.dec_blocks = nn.ModuleList([
+            DecoderBlock(dec_embed_dim, dec_num_heads, mlp_ratio=mlp_ratio, qkv_bias=True, norm_layer=norm_layer, norm_mem=norm_im2_in_dec, rope=self.rope)
+            for i in range(dec_depth)])
+        # final norm layer 
+        self.dec_norm = norm_layer(dec_embed_dim)
+        
+    def _set_prediction_head(self, dec_embed_dim, patch_size):
+         self.prediction_head = nn.Linear(dec_embed_dim, patch_size**2 * 3, bias=True)
+        
+        
+    def initialize_weights(self):
+        # patch embed 
+        self.patch_embed._init_weights()
+        # mask tokens
+        if self.mask_token is not None: torch.nn.init.normal_(self.mask_token, std=.02)
+        # linears and layer norms
+        self.apply(self._init_weights)
+
+    def _init_weights(self, m):
+        if isinstance(m, nn.Linear):
+            # we use xavier_uniform following official JAX ViT:
+            torch.nn.init.xavier_uniform_(m.weight)
+            if isinstance(m, nn.Linear) and m.bias is not None:
+                nn.init.constant_(m.bias, 0)
+        elif isinstance(m, nn.LayerNorm):
+            nn.init.constant_(m.bias, 0)
+            nn.init.constant_(m.weight, 1.0)
+            
+    def _encode_image(self, image, do_mask=False, return_all_blocks=False):
+        """
+        image has B x 3 x img_size x img_size 
+        do_mask: whether to perform masking or not
+        return_all_blocks: if True, return the features at the end of every block 
+                           instead of just the features from the last block (eg for some prediction heads)
+        """
+        # embed the image into patches  (x has size B x Npatches x C) 
+        # and get position if each return patch (pos has size B x Npatches x 2)
+        x, pos = self.patch_embed(image)              
+        # add positional embedding without cls token  
+        if self.enc_pos_embed is not None: 
+            x = x + self.enc_pos_embed[None,...]
+        # apply masking 
+        B,N,C = x.size()
+        if do_mask:
+            masks = self.mask_generator(x)
+            x = x[~masks].view(B, -1, C)
+            posvis = pos[~masks].view(B, -1, 2)
+        else:
+            B,N,C = x.size()
+            masks = torch.zeros((B,N), dtype=bool)
+            posvis = pos
+        # now apply the transformer encoder and normalization        
+        if return_all_blocks:
+            out = []
+            for blk in self.enc_blocks:
+                x = blk(x, posvis)
+                out.append(x)
+            out[-1] = self.enc_norm(out[-1])
+            return out, pos, masks
+        else:
+            for blk in self.enc_blocks:
+                x = blk(x, posvis)
+            x = self.enc_norm(x)
+            return x, pos, masks
+ 
+    def _decoder(self, feat1, pos1, masks1, feat2, pos2, return_all_blocks=False):
+        """
+        return_all_blocks: if True, return the features at the end of every block 
+                           instead of just the features from the last block (eg for some prediction heads)
+                           
+        masks1 can be None => assume image1 fully visible 
+        """
+        # encoder to decoder layer 
+        visf1 = self.decoder_embed(feat1)
+        f2 = self.decoder_embed(feat2)
+        # append masked tokens to the sequence
+        B,Nenc,C = visf1.size()
+        if masks1 is None: # downstreams
+            f1_ = visf1
+        else: # pretraining 
+            Ntotal = masks1.size(1)
+            f1_ = self.mask_token.repeat(B, Ntotal, 1).to(dtype=visf1.dtype)
+            f1_[~masks1] = visf1.view(B * Nenc, C)
+        # add positional embedding
+        if self.dec_pos_embed is not None:
+            f1_ = f1_ + self.dec_pos_embed
+            f2 = f2 + self.dec_pos_embed
+        # apply Transformer blocks
+        out = f1_
+        out2 = f2 
+        if return_all_blocks:
+            _out, out = out, []
+            for blk in self.dec_blocks:
+                _out, out2 = blk(_out, out2, pos1, pos2)
+                out.append(_out)
+            out[-1] = self.dec_norm(out[-1])
+        else:
+            for blk in self.dec_blocks:
+                out, out2 = blk(out, out2, pos1, pos2)
+            out = self.dec_norm(out)
+        return out
+
+    def patchify(self, imgs):
+        """
+        imgs: (B, 3, H, W)
+        x: (B, L, patch_size**2 *3)
+        """
+        p = self.patch_embed.patch_size[0]
+        assert imgs.shape[2] == imgs.shape[3] and imgs.shape[2] % p == 0
+
+        h = w = imgs.shape[2] // p
+        x = imgs.reshape(shape=(imgs.shape[0], 3, h, p, w, p))
+        x = torch.einsum('nchpwq->nhwpqc', x)
+        x = x.reshape(shape=(imgs.shape[0], h * w, p**2 * 3))
+        
+        return x
+
+    def unpatchify(self, x, channels=3):
+        """
+        x: (N, L, patch_size**2 *channels)
+        imgs: (N, 3, H, W)
+        """
+        patch_size = self.patch_embed.patch_size[0]
+        h = w = int(x.shape[1]**.5)
+        assert h * w == x.shape[1]
+        x = x.reshape(shape=(x.shape[0], h, w, patch_size, patch_size, channels))
+        x = torch.einsum('nhwpqc->nchpwq', x)
+        imgs = x.reshape(shape=(x.shape[0], channels, h * patch_size, h * patch_size))
+        return imgs
+
+    def forward(self, img1, img2):
+        """
+        img1: tensor of size B x 3 x img_size x img_size
+        img2: tensor of size B x 3 x img_size x img_size
+        
+        out will be    B x N x (3*patch_size*patch_size)
+        masks are also returned as B x N just in case 
+        """
+        # encoder of the masked first image 
+        feat1, pos1, mask1 = self._encode_image(img1, do_mask=True)
+        # encoder of the second image 
+        feat2, pos2, _ = self._encode_image(img2, do_mask=False)
+        # decoder 
+        decfeat = self._decoder(feat1, pos1, mask1, feat2, pos2)
+        # prediction head 
+        out = self.prediction_head(decfeat)
+        # get target
+        target = self.patchify(img1)
+        return out, mask1, target

croco/models/croco_downstream.py

@@ -0,0 +1,122 @@
+# Copyright (C) 2022-present Naver Corporation. All rights reserved.
+# Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+
+# --------------------------------------------------------
+# CroCo model for downstream tasks
+# --------------------------------------------------------
+
+import torch
+
+from .croco import CroCoNet
+
+
+def croco_args_from_ckpt(ckpt):
+    if 'croco_kwargs' in ckpt: # CroCo v2 released models
+        return ckpt['croco_kwargs']
+    elif 'args' in ckpt and hasattr(ckpt['args'], 'model'): # pretrained using the official code release
+        s = ckpt['args'].model # eg "CroCoNet(enc_embed_dim=1024, enc_num_heads=16, enc_depth=24)"
+        assert s.startswith('CroCoNet(')
+        return eval('dict'+s[len('CroCoNet'):]) # transform it into the string of a dictionary and evaluate it
+    else: # CroCo v1 released models
+        return dict()
+
+class CroCoDownstreamMonocularEncoder(CroCoNet):
+
+    def __init__(self,
+                 head,
+                 **kwargs):
+        """ Build network for monocular downstream task, only using the encoder.
+        It takes an extra argument head, that is called with the features 
+          and a dictionary img_info containing 'width' and 'height' keys
+        The head is setup with the croconet arguments in this init function
+        NOTE: It works by *calling super().__init__() but with redefined setters
+        
+        """
+        super(CroCoDownstreamMonocularEncoder, self).__init__(**kwargs)
+        head.setup(self)
+        self.head = head
+
+    def _set_mask_generator(self, *args, **kwargs):
+        """ No mask generator """
+        return
+
+    def _set_mask_token(self, *args, **kwargs):
+        """ No mask token """
+        self.mask_token = None
+        return
+
+    def _set_decoder(self, *args, **kwargs):
+        """ No decoder """
+        return
+
+    def _set_prediction_head(self, *args, **kwargs):
+        """ No 'prediction head' for downstream tasks."""
+        return
+
+    def forward(self, img):
+        """
+        img if of size batch_size x 3 x h x w
+        """
+        B, C, H, W = img.size()
+        img_info = {'height': H, 'width': W}
+        need_all_layers = hasattr(self.head, 'return_all_blocks') and self.head.return_all_blocks
+        out, _, _ = self._encode_image(img, do_mask=False, return_all_blocks=need_all_layers)
+        return self.head(out, img_info)
+        
+        
+class CroCoDownstreamBinocular(CroCoNet):
+
+    def __init__(self,
+                 head,
+                 **kwargs):
+        """ Build network for binocular downstream task
+        It takes an extra argument head, that is called with the features 
+          and a dictionary img_info containing 'width' and 'height' keys
+        The head is setup with the croconet arguments in this init function
+        """
+        super(CroCoDownstreamBinocular, self).__init__(**kwargs)
+        head.setup(self)
+        self.head = head
+
+    def _set_mask_generator(self, *args, **kwargs):
+        """ No mask generator """
+        return
+
+    def _set_mask_token(self, *args, **kwargs):
+        """ No mask token """
+        self.mask_token = None
+        return
+
+    def _set_prediction_head(self, *args, **kwargs):
+        """ No prediction head for downstream tasks, define your own head """
+        return
+        
+    def encode_image_pairs(self, img1, img2, return_all_blocks=False):
+        """ run encoder for a pair of images
+            it is actually ~5% faster to concatenate the images along the batch dimension 
+             than to encode them separately
+        """
+        ## the two commented lines below is the naive version with separate encoding
+        #out, pos, _ = self._encode_image(img1, do_mask=False, return_all_blocks=return_all_blocks)
+        #out2, pos2, _ = self._encode_image(img2, do_mask=False, return_all_blocks=False)
+        ## and now the faster version
+        out, pos, _ = self._encode_image( torch.cat( (img1,img2), dim=0), do_mask=False, return_all_blocks=return_all_blocks )
+        if return_all_blocks:
+            out,out2 = list(map(list, zip(*[o.chunk(2, dim=0) for o in out])))
+            out2 = out2[-1]
+        else:
+            out,out2 = out.chunk(2, dim=0)
+        pos,pos2 = pos.chunk(2, dim=0)            
+        return out, out2, pos, pos2
+
+    def forward(self, img1, img2):
+        B, C, H, W = img1.size()
+        img_info = {'height': H, 'width': W}
+        return_all_blocks = hasattr(self.head, 'return_all_blocks') and self.head.return_all_blocks
+        out, out2, pos, pos2 = self.encode_image_pairs(img1, img2, return_all_blocks=return_all_blocks)
+        if return_all_blocks:
+            decout = self._decoder(out[-1], pos, None, out2, pos2, return_all_blocks=return_all_blocks)
+            decout = out+decout
+        else:
+            decout = self._decoder(out, pos, None, out2, pos2, return_all_blocks=return_all_blocks)
+        return self.head(decout, img_info)

croco/models/curope/__init__.py

@@ -0,0 +1,4 @@
+# Copyright (C) 2022-present Naver Corporation. All rights reserved.
+# Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+
+from .curope2d import cuRoPE2D

croco/models/curope/build/temp.linux-x86_64-cpython-311/.ninja_log

@@ -0,0 +1,3 @@
+# ninja log v5
+0	12275	1728651353555686417	/home/lipeng/ljh_code/Video_Depth_CVPR2025-main/dust3r_train/croco/models/curope/build/temp.linux-x86_64-cpython-311/curope.o	46197fb3a2f9e5ab
+0	156505	1728651497679855739	/home/lipeng/ljh_code/Video_Depth_CVPR2025-main/dust3r_train/croco/models/curope/build/temp.linux-x86_64-cpython-311/kernels.o	5a65d5447ccdcd9c

croco/models/curope/build/temp.linux-x86_64-cpython-311/build.ninja

@@ -0,0 +1,33 @@
+ninja_required_version = 1.3
+cxx = c++
+nvcc = /usr/local/cuda/bin/nvcc
+
+cflags = -pthread -B /home/lipeng/miniconda3/envs/dust3r/compiler_compat -DNDEBUG -fwrapv -O2 -Wall -fPIC -O2 -isystem /home/lipeng/miniconda3/envs/dust3r/include -fPIC -O2 -isystem /home/lipeng/miniconda3/envs/dust3r/include -fPIC -I/home/lipeng/miniconda3/envs/dust3r/lib/python3.11/site-packages/torch/include -I/home/lipeng/miniconda3/envs/dust3r/lib/python3.11/site-packages/torch/include/torch/csrc/api/include -I/home/lipeng/miniconda3/envs/dust3r/lib/python3.11/site-packages/torch/include/TH -I/home/lipeng/miniconda3/envs/dust3r/lib/python3.11/site-packages/torch/include/THC -I/usr/local/cuda/include -I/home/lipeng/miniconda3/envs/dust3r/include/python3.11 -c
+post_cflags = -O3 -DTORCH_API_INCLUDE_EXTENSION_H '-DPYBIND11_COMPILER_TYPE="_gcc"' '-DPYBIND11_STDLIB="_libstdcpp"' '-DPYBIND11_BUILD_ABI="_cxxabi1011"' -DTORCH_EXTENSION_NAME=curope -D_GLIBCXX_USE_CXX11_ABI=0 -std=c++17
+cuda_cflags = -I/home/lipeng/miniconda3/envs/dust3r/lib/python3.11/site-packages/torch/include -I/home/lipeng/miniconda3/envs/dust3r/lib/python3.11/site-packages/torch/include/torch/csrc/api/include -I/home/lipeng/miniconda3/envs/dust3r/lib/python3.11/site-packages/torch/include/TH -I/home/lipeng/miniconda3/envs/dust3r/lib/python3.11/site-packages/torch/include/THC -I/usr/local/cuda/include -I/home/lipeng/miniconda3/envs/dust3r/include/python3.11 -c
+cuda_post_cflags = -D__CUDA_NO_HALF_OPERATORS__ -D__CUDA_NO_HALF_CONVERSIONS__ -D__CUDA_NO_BFLOAT16_CONVERSIONS__ -D__CUDA_NO_HALF2_OPERATORS__ --expt-relaxed-constexpr --compiler-options ''"'"'-fPIC'"'"'' -O3 --ptxas-options=-v --use_fast_math -gencode arch=compute_50,code=sm_50 -gencode arch=compute_60,code=sm_60 -gencode arch=compute_61,code=sm_61 -gencode arch=compute_70,code=sm_70 -gencode arch=compute_75,code=sm_75 -gencode arch=compute_80,code=sm_80 -gencode arch=compute_86,code=sm_86 -gencode arch=compute_90,code=sm_90 -DTORCH_API_INCLUDE_EXTENSION_H '-DPYBIND11_COMPILER_TYPE="_gcc"' '-DPYBIND11_STDLIB="_libstdcpp"' '-DPYBIND11_BUILD_ABI="_cxxabi1011"' -DTORCH_EXTENSION_NAME=curope -D_GLIBCXX_USE_CXX11_ABI=0 -std=c++17
+cuda_dlink_post_cflags = 
+ldflags = 
+
+rule compile
+  command = $cxx -MMD -MF $out.d $cflags -c $in -o $out $post_cflags
+  depfile = $out.d
+  deps = gcc
+
+rule cuda_compile
+  depfile = $out.d
+  deps = gcc
+  command = $nvcc --generate-dependencies-with-compile --dependency-output $out.d $cuda_cflags -c $in -o $out $cuda_post_cflags
+
+
+
+
+
+build /home/lipeng/ljh_code/Video_Depth_CVPR2025-main/dust3r_train/croco/models/curope/build/temp.linux-x86_64-cpython-311/curope.o: compile /home/lipeng/ljh_code/Video_Depth_CVPR2025-main/dust3r_train/croco/models/curope/curope.cpp
+build /home/lipeng/ljh_code/Video_Depth_CVPR2025-main/dust3r_train/croco/models/curope/build/temp.linux-x86_64-cpython-311/kernels.o: cuda_compile /home/lipeng/ljh_code/Video_Depth_CVPR2025-main/dust3r_train/croco/models/curope/kernels.cu
+
+
+
+
+
+

croco/models/curope/curope.cpp

@@ -0,0 +1,69 @@
+/* 
+  Copyright (C) 2022-present Naver Corporation. All rights reserved.
+  Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+*/
+
+#include <torch/extension.h>
+
+// forward declaration
+void rope_2d_cuda( torch::Tensor tokens, const torch::Tensor pos, const float base, const float fwd );
+
+void rope_2d_cpu( torch::Tensor tokens, const torch::Tensor positions, const float base, const float fwd )
+{
+    const int B = tokens.size(0);
+    const int N = tokens.size(1);
+    const int H = tokens.size(2);
+    const int D = tokens.size(3) / 4;
+
+    auto tok = tokens.accessor<float, 4>();
+    auto pos = positions.accessor<int64_t, 3>();
+
+    for (int b = 0; b < B; b++) {
+      for (int x = 0; x < 2; x++) { // y and then x (2d)
+        for (int n = 0; n < N; n++) {
+        
+            // grab the token position
+            const int p = pos[b][n][x];
+
+            for (int h = 0; h < H; h++) {
+                for (int d = 0; d < D; d++) {
+                    // grab the two values
+                    float u = tok[b][n][h][d+0+x*2*D];
+                    float v = tok[b][n][h][d+D+x*2*D];
+
+                    // grab the cos,sin
+                    const float inv_freq = fwd * p / powf(base, d/float(D));
+                    float c = cosf(inv_freq);
+                    float s = sinf(inv_freq);
+
+                    // write the result
+                    tok[b][n][h][d+0+x*2*D] = u*c - v*s;
+                    tok[b][n][h][d+D+x*2*D] = v*c + u*s;
+                }
+            }
+        }
+      }
+    }
+}
+
+void rope_2d( torch::Tensor tokens,     // B,N,H,D
+        const torch::Tensor positions,  // B,N,2
+        const float base, 
+        const float fwd )
+{
+    TORCH_CHECK(tokens.dim() == 4, "tokens must have 4 dimensions");
+    TORCH_CHECK(positions.dim() == 3, "positions must have 3 dimensions");
+    TORCH_CHECK(tokens.size(0) == positions.size(0), "batch size differs between tokens & positions");
+    TORCH_CHECK(tokens.size(1) == positions.size(1), "seq_length differs between tokens & positions");
+    TORCH_CHECK(positions.size(2) == 2, "positions.shape[2] must be equal to 2");
+    TORCH_CHECK(tokens.is_cuda() == positions.is_cuda(), "tokens and positions are not on the same device" );
+
+    if (tokens.is_cuda())
+        rope_2d_cuda( tokens, positions, base, fwd );
+    else
+        rope_2d_cpu( tokens, positions, base, fwd );
+}
+
+PYBIND11_MODULE(TORCH_EXTENSION_NAME, m) {
+  m.def("rope_2d", &rope_2d, "RoPE 2d forward/backward");
+}

croco/models/curope/curope2d.py

@@ -0,0 +1,40 @@
+# Copyright (C) 2022-present Naver Corporation. All rights reserved.
+# Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+
+import torch
+
+try:
+    import curope as _kernels # run `python setup.py install`
+except ModuleNotFoundError:
+    from . import curope as _kernels # run `python setup.py build_ext --inplace`
+
+
+class cuRoPE2D_func (torch.autograd.Function):
+
+    @staticmethod
+    def forward(ctx, tokens, positions, base, F0=1):
+        ctx.save_for_backward(positions)
+        ctx.saved_base = base
+        ctx.saved_F0 = F0
+        # tokens = tokens.clone() # uncomment this if inplace doesn't work
+        _kernels.rope_2d( tokens, positions, base, F0 )
+        ctx.mark_dirty(tokens)
+        return tokens
+
+    @staticmethod
+    def backward(ctx, grad_res):
+        positions, base, F0 = ctx.saved_tensors[0], ctx.saved_base, ctx.saved_F0
+        _kernels.rope_2d( grad_res, positions, base, -F0 )
+        ctx.mark_dirty(grad_res)
+        return grad_res, None, None, None
+
+
+class cuRoPE2D(torch.nn.Module):
+    def __init__(self, freq=100.0, F0=1.0):
+        super().__init__()
+        self.base = freq 
+        self.F0 = F0
+
+    def forward(self, tokens, positions): 
+        cuRoPE2D_func.apply( tokens.transpose(1,2), positions, self.base, self.F0 )
+        return tokens

croco/models/curope/kernels.cu

@@ -0,0 +1,108 @@
+/* 
+  Copyright (C) 2022-present Naver Corporation. All rights reserved.
+  Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+*/
+
+#include <torch/extension.h>
+#include <cuda.h>
+#include <cuda_runtime.h>
+#include <vector>
+
+#define CHECK_CUDA(tensor) {\
+    TORCH_CHECK((tensor).is_cuda(), #tensor " is not in cuda memory"); \
+    TORCH_CHECK((tensor).is_contiguous(), #tensor " is not contiguous"); }
+void CHECK_KERNEL() {auto error = cudaGetLastError(); TORCH_CHECK( error == cudaSuccess, cudaGetErrorString(error));}
+
+
+template < typename scalar_t  >
+__global__ void rope_2d_cuda_kernel( 
+        //scalar_t* __restrict__ tokens, 
+        torch::PackedTensorAccessor32<scalar_t,4,torch::RestrictPtrTraits> tokens,
+        const int64_t* __restrict__ pos, 
+        const float base, 
+        const float fwd )
+        // const int N, const int H, const int D )
+{
+    // tokens shape = (B, N, H, D)
+    const int N = tokens.size(1);
+    const int H = tokens.size(2);
+    const int D = tokens.size(3);
+    
+    // each block update a single token, for all heads
+    // each thread takes care of a single output
+    extern __shared__ float shared[];
+    float* shared_inv_freq = shared + D;
+
+    const int b = blockIdx.x / N;
+    const int n = blockIdx.x % N;
+
+    const int Q = D / 4; 
+    // one token = [0..Q : Q..2Q : 2Q..3Q : 3Q..D]
+    //              u_Y     v_Y     u_X      v_X
+
+    // shared memory: first, compute inv_freq
+    if (threadIdx.x < Q)
+        shared_inv_freq[threadIdx.x] = fwd / powf(base, threadIdx.x/float(Q));
+    __syncthreads();
+
+    // start of X or Y part
+    const int X = threadIdx.x < D/2 ? 0 : 1; 
+    const int m = (X*D/2) + (threadIdx.x % Q);   // index of u_Y or u_X
+
+    // grab the cos,sin appropriate for me
+    const float freq = pos[blockIdx.x*2+X] * shared_inv_freq[threadIdx.x % Q];
+    const float cos = cosf(freq);
+    const float sin = sinf(freq);
+    /*
+    float* shared_cos_sin = shared + D + D/4;
+    if ((threadIdx.x % (D/2)) < Q)
+        shared_cos_sin[m+0] = cosf(freq);
+    else
+        shared_cos_sin[m+Q] = sinf(freq);
+    __syncthreads();
+    const float cos = shared_cos_sin[m+0];
+    const float sin = shared_cos_sin[m+Q];
+    */
+
+    for (int h = 0; h < H; h++)
+    {
+        // then, load all the token for this head in shared memory
+        shared[threadIdx.x] = tokens[b][n][h][threadIdx.x];
+        __syncthreads();
+
+        const float u = shared[m];
+        const float v = shared[m+Q];
+        
+        // write output
+        if ((threadIdx.x % (D/2)) < Q)
+            tokens[b][n][h][threadIdx.x] = u*cos - v*sin;
+        else
+            tokens[b][n][h][threadIdx.x] = v*cos + u*sin;
+    }
+}
+
+void rope_2d_cuda( torch::Tensor tokens, const torch::Tensor pos, const float base, const float fwd ) 
+{
+    const int B = tokens.size(0); // batch size
+    const int N = tokens.size(1); // sequence length
+    const int H = tokens.size(2); // number of heads
+    const int D = tokens.size(3); // dimension per head
+
+    TORCH_CHECK(tokens.stride(3) == 1 && tokens.stride(2) == D, "tokens are not contiguous");
+    TORCH_CHECK(pos.is_contiguous(), "positions are not contiguous");
+    TORCH_CHECK(pos.size(0) == B && pos.size(1) == N && pos.size(2) == 2, "bad pos.shape");
+    TORCH_CHECK(D % 4 == 0, "token dim must be multiple of 4");
+
+    // one block for each layer, one thread per local-max
+    const int THREADS_PER_BLOCK = D;
+    const int N_BLOCKS = B * N; // each block takes care of H*D values
+    const int SHARED_MEM = sizeof(float) * (D + D/4);
+
+    AT_DISPATCH_FLOATING_TYPES_AND_HALF(tokens.type(), "rope_2d_cuda", ([&] {
+        rope_2d_cuda_kernel<scalar_t> <<<N_BLOCKS, THREADS_PER_BLOCK, SHARED_MEM>>> (
+            //tokens.data_ptr<scalar_t>(), 
+            tokens.packed_accessor32<scalar_t,4,torch::RestrictPtrTraits>(),
+            pos.data_ptr<int64_t>(), 
+            base, fwd); //, N, H, D );
+    }));
+}

croco/models/curope/setup.py

@@ -0,0 +1,34 @@
+# Copyright (C) 2022-present Naver Corporation. All rights reserved.
+# Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+
+from setuptools import setup
+from torch import cuda
+from torch.utils.cpp_extension import BuildExtension, CUDAExtension
+
+# compile for all possible CUDA architectures
+all_cuda_archs = cuda.get_gencode_flags().replace('compute=','arch=').split()
+# alternatively, you can list cuda archs that you want, eg:
+# all_cuda_archs = [
+    # '-gencode', 'arch=compute_70,code=sm_70',
+    # '-gencode', 'arch=compute_75,code=sm_75',
+    # '-gencode', 'arch=compute_80,code=sm_80',
+    # '-gencode', 'arch=compute_86,code=sm_86'
+# ]
+
+setup(
+    name = 'curope',
+    ext_modules = [
+        CUDAExtension(
+                name='curope',
+                sources=[
+                    "curope.cpp",
+                    "kernels.cu",
+                ],
+                extra_compile_args = dict(
+                    nvcc=['-O3','--ptxas-options=-v',"--use_fast_math"]+all_cuda_archs, 
+                    cxx=['-O3'])
+                )
+    ],
+    cmdclass = {
+        'build_ext': BuildExtension
+    })

croco/models/dpt_block.py

@@ -0,0 +1,457 @@
+# Copyright (C) 2022-present Naver Corporation. All rights reserved.
+# Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+
+# --------------------------------------------------------
+# DPT head for ViTs
+# --------------------------------------------------------
+# References: 
+# https://github.com/isl-org/DPT
+# https://github.com/EPFL-VILAB/MultiMAE/blob/main/multimae/output_adapters.py
+
+import torch
+import torch.nn as nn
+import torch.nn.functional as F
+from einops import rearrange, repeat
+from typing import Union, Tuple, Iterable, List, Optional, Dict
+
+def pair(t):
+    return t if isinstance(t, tuple) else (t, t)
+
+def make_scratch(in_shape, out_shape, groups=1, expand=False):
+    scratch = nn.Module()
+
+    out_shape1 = out_shape
+    out_shape2 = out_shape
+    out_shape3 = out_shape
+    out_shape4 = out_shape
+    if expand == True:
+        out_shape1 = out_shape
+        out_shape2 = out_shape * 2
+        out_shape3 = out_shape * 4
+        out_shape4 = out_shape * 8
+
+    scratch.layer1_rn = nn.Conv2d(
+        in_shape[0],
+        out_shape1,
+        kernel_size=3,
+        stride=1,
+        padding=1,
+        bias=False,
+        groups=groups,
+    )
+    scratch.layer2_rn = nn.Conv2d(
+        in_shape[1],
+        out_shape2,
+        kernel_size=3,
+        stride=1,
+        padding=1,
+        bias=False,
+        groups=groups,
+    )
+    scratch.layer3_rn = nn.Conv2d(
+        in_shape[2],
+        out_shape3,
+        kernel_size=3,
+        stride=1,
+        padding=1,
+        bias=False,
+        groups=groups,
+    )
+    scratch.layer4_rn = nn.Conv2d(
+        in_shape[3],
+        out_shape4,
+        kernel_size=3,
+        stride=1,
+        padding=1,
+        bias=False,
+        groups=groups,
+    )
+
+    scratch.layer_rn = nn.ModuleList([
+        scratch.layer1_rn,
+        scratch.layer2_rn,
+        scratch.layer3_rn,
+        scratch.layer4_rn,
+    ])
+
+    return scratch
+
+class ResidualConvUnit_custom(nn.Module):
+    """Residual convolution module."""
+
+    def __init__(self, features, activation, bn):
+        """Init.
+        Args:
+            features (int): number of features
+        """
+        super().__init__()
+
+        self.bn = bn
+
+        self.groups = 1
+
+        self.conv1 = nn.Conv2d(
+            features,
+            features,
+            kernel_size=3,
+            stride=1,
+            padding=1,
+            bias=not self.bn,
+            groups=self.groups,
+        )
+
+        self.conv2 = nn.Conv2d(
+            features,
+            features,
+            kernel_size=3,
+            stride=1,
+            padding=1,
+            bias=not self.bn,
+            groups=self.groups,
+        )
+
+        if self.bn == True:
+            self.bn1 = nn.BatchNorm2d(features)
+            self.bn2 = nn.BatchNorm2d(features)
+
+        self.activation = activation
+
+        self.skip_add = nn.quantized.FloatFunctional()
+
+    def forward(self, x):
+        """Forward pass.
+        Args:
+            x (tensor): input
+        Returns:
+            tensor: output
+        """
+
+        out = self.activation(x)
+        out = self.conv1(out)
+        if self.bn == True:
+            out = self.bn1(out)
+
+        out = self.activation(out)
+        out = self.conv2(out)
+        if self.bn == True:
+            out = self.bn2(out)
+
+        if self.groups > 1:
+            out = self.conv_merge(out)
+
+        return self.skip_add.add(out, x)
+
+class FeatureFusionBlock_custom(nn.Module):
+    """Feature fusion block."""
+
+    def __init__(
+        self,
+        features,
+        activation,
+        deconv=False,
+        bn=False,
+        expand=False,
+        align_corners=True,
+        width_ratio=1,
+    ):
+        """Init.
+        Args:
+            features (int): number of features
+        """
+        super(FeatureFusionBlock_custom, self).__init__()
+        self.width_ratio = width_ratio
+
+        self.deconv = deconv
+        self.align_corners = align_corners
+
+        self.groups = 1
+
+        self.expand = expand
+        out_features = features
+        if self.expand == True:
+            out_features = features // 2
+
+        self.out_conv = nn.Conv2d(
+            features,
+            out_features,
+            kernel_size=1,
+            stride=1,
+            padding=0,
+            bias=True,
+            groups=1,
+        )
+
+        self.resConfUnit1 = ResidualConvUnit_custom(features, activation, bn)
+        self.resConfUnit2 = ResidualConvUnit_custom(features, activation, bn)
+
+        self.skip_add = nn.quantized.FloatFunctional()
+
+    def forward(self, *xs):
+        """Forward pass.
+        Returns:
+            tensor: output
+        """
+        output = xs[0]
+
+        if len(xs) == 2:
+            res = self.resConfUnit1(xs[1])
+            if self.width_ratio != 1:
+                res = F.interpolate(res, size=(output.shape[2], output.shape[3]), mode='bilinear')
+
+            output = self.skip_add.add(output, res)
+            # output += res
+
+        output = self.resConfUnit2(output)
+
+        if self.width_ratio != 1:
+            # and output.shape[3] < self.width_ratio * output.shape[2]
+            #size=(image.shape[])
+            if (output.shape[3] / output.shape[2]) < (2 / 3) * self.width_ratio:
+                shape = 3 * output.shape[3]
+            else:
+                shape = int(self.width_ratio * 2 * output.shape[2])
+            output  = F.interpolate(output, size=(2* output.shape[2], shape), mode='bilinear')
+        else:
+            output = nn.functional.interpolate(output, scale_factor=2,
+                    mode="bilinear", align_corners=self.align_corners)
+        output = self.out_conv(output)
+        return output
+
+def make_fusion_block(features, use_bn, width_ratio=1):
+    return FeatureFusionBlock_custom(
+        features,
+        nn.ReLU(False),
+        deconv=False,
+        bn=use_bn,
+        expand=False,
+        align_corners=True,
+        width_ratio=width_ratio,
+    )
+
+class Interpolate(nn.Module):
+    """Interpolation module."""
+
+    def __init__(self, scale_factor, mode, align_corners=False):
+        """Init.
+        Args:
+            scale_factor (float): scaling
+            mode (str): interpolation mode
+        """
+        super(Interpolate, self).__init__()
+
+        self.interp = nn.functional.interpolate
+        self.scale_factor = scale_factor
+        self.mode = mode
+        self.align_corners = align_corners
+
+    def forward(self, x):
+        """Forward pass.
+        Args:
+            x (tensor): input
+        Returns:
+            tensor: interpolated data
+        """
+
+        x = self.interp(
+            x,
+            scale_factor=self.scale_factor,
+            mode=self.mode,
+            align_corners=self.align_corners,
+        )
+
+        return x
+
+class DPTOutputAdapter(nn.Module):
+    """DPT output adapter.
+
+    :param num_cahnnels: Number of output channels
+    :param stride_level: tride level compared to the full-sized image.
+        E.g. 4 for 1/4th the size of the image.
+    :param patch_size_full: Int or tuple of the patch size over the full image size.
+        Patch size for smaller inputs will be computed accordingly.
+    :param hooks: Index of intermediate layers
+    :param layer_dims: Dimension of intermediate layers
+    :param feature_dim: Feature dimension
+    :param last_dim: out_channels/in_channels for the last two Conv2d when head_type == regression
+    :param use_bn: If set to True, activates batch norm
+    :param dim_tokens_enc:  Dimension of tokens coming from encoder
+    """
+
+    def __init__(self,
+                 num_channels: int = 1,
+                 stride_level: int = 1,
+                 patch_size: Union[int, Tuple[int, int]] = 16,
+                 main_tasks: Iterable[str] = ('rgb',),
+                 hooks: List[int] = [2, 5, 8, 11],
+                 layer_dims: List[int] = [96, 192, 384, 768],
+                 feature_dim: int = 256,
+                 last_dim: int = 32,
+                 use_bn: bool = False,
+                 dim_tokens_enc: Optional[int] = None,
+                 head_type: str = 'regression',
+                 output_width_ratio=1,
+                 **kwargs):
+        super().__init__()
+        self.num_channels = num_channels
+        self.stride_level = stride_level
+        self.patch_size = pair(patch_size)
+        self.main_tasks = main_tasks
+        self.hooks = hooks
+        self.layer_dims = layer_dims
+        self.feature_dim = feature_dim
+        self.dim_tokens_enc = dim_tokens_enc * len(self.main_tasks) if dim_tokens_enc is not None else None
+        self.head_type = head_type
+
+        # Actual patch height and width, taking into account stride of input
+        self.P_H = max(1, self.patch_size[0] // stride_level)
+        self.P_W = max(1, self.patch_size[1] // stride_level)
+
+        self.scratch = make_scratch(layer_dims, feature_dim, groups=1, expand=False)
+
+        self.scratch.refinenet1 = make_fusion_block(feature_dim, use_bn, output_width_ratio)
+        self.scratch.refinenet2 = make_fusion_block(feature_dim, use_bn, output_width_ratio)
+        self.scratch.refinenet3 = make_fusion_block(feature_dim, use_bn, output_width_ratio)
+        self.scratch.refinenet4 = make_fusion_block(feature_dim, use_bn, output_width_ratio)
+        # self.mask_head = nn.Sequential(
+        #         nn.Conv2d(feature_dim, feature_dim, kernel_size=3, padding=1, bias=False),
+        #         nn.BatchNorm2d(feature_dim) if use_bn else nn.Identity(),
+        #         nn.ReLU(True),
+        #         nn.Dropout(0.1, False),
+        #         nn.Conv2d(feature_dim, 1, kernel_size=1),
+        #         Interpolate(scale_factor=2, mode="bilinear", align_corners=True),
+        #     )
+        if self.head_type == 'regression':
+            # The "DPTDepthModel" head
+            self.head = nn.Sequential(
+                nn.Conv2d(feature_dim, feature_dim // 2, kernel_size=3, stride=1, padding=1),
+                Interpolate(scale_factor=2, mode="bilinear", align_corners=True),
+                nn.Conv2d(feature_dim // 2, last_dim, kernel_size=3, stride=1, padding=1),
+                nn.ReLU(True),
+                nn.Conv2d(last_dim, self.num_channels, kernel_size=1, stride=1, padding=0)
+            )
+        elif self.head_type == 'semseg':
+            # The "DPTSegmentationModel" head
+            self.head = nn.Sequential(
+                nn.Conv2d(feature_dim, feature_dim, kernel_size=3, padding=1, bias=False),
+                nn.BatchNorm2d(feature_dim) if use_bn else nn.Identity(),
+                nn.ReLU(True),
+                nn.Dropout(0.1, False),
+                nn.Conv2d(feature_dim, self.num_channels, kernel_size=1),
+                Interpolate(scale_factor=2, mode="bilinear", align_corners=True),
+            )
+        else:
+            raise ValueError('DPT head_type must be "regression" or "semseg".')
+
+        if self.dim_tokens_enc is not None:
+            self.init(dim_tokens_enc=dim_tokens_enc)
+
+    def init(self, dim_tokens_enc=768):
+        """
+        Initialize parts of decoder that are dependent on dimension of encoder tokens.
+        Should be called when setting up MultiMAE.
+
+        :param dim_tokens_enc: Dimension of tokens coming from encoder
+        """
+        #print(dim_tokens_enc)
+
+        # Set up activation postprocessing layers
+        if isinstance(dim_tokens_enc, int):
+            dim_tokens_enc = 4 * [dim_tokens_enc]
+
+        self.dim_tokens_enc = [dt * len(self.main_tasks) for dt in dim_tokens_enc]
+
+        self.act_1_postprocess = nn.Sequential(
+            nn.Conv2d(
+                in_channels=self.dim_tokens_enc[0],
+                out_channels=self.layer_dims[0],
+                kernel_size=1, stride=1, padding=0,
+            ),
+            nn.ConvTranspose2d(
+                in_channels=self.layer_dims[0],
+                out_channels=self.layer_dims[0],
+                kernel_size=4, stride=4, padding=0,
+                bias=True, dilation=1, groups=1,
+            )
+        )
+
+        self.act_2_postprocess = nn.Sequential(
+            nn.Conv2d(
+                in_channels=self.dim_tokens_enc[1],
+                out_channels=self.layer_dims[1],
+                kernel_size=1, stride=1, padding=0,
+            ),
+            nn.ConvTranspose2d(
+                in_channels=self.layer_dims[1],
+                out_channels=self.layer_dims[1],
+                kernel_size=2, stride=2, padding=0,
+                bias=True, dilation=1, groups=1,
+            )
+        )
+
+        self.act_3_postprocess = nn.Sequential(
+            nn.Conv2d(
+                in_channels=self.dim_tokens_enc[2],
+                out_channels=self.layer_dims[2],
+                kernel_size=1, stride=1, padding=0,
+            )
+        )
+
+        self.act_4_postprocess = nn.Sequential(
+            nn.Conv2d(
+                in_channels=self.dim_tokens_enc[3],
+                out_channels=self.layer_dims[3],
+                kernel_size=1, stride=1, padding=0,
+            ),
+            nn.Conv2d(
+                in_channels=self.layer_dims[3],
+                out_channels=self.layer_dims[3],
+                kernel_size=3, stride=2, padding=1,
+            )
+        )
+
+        self.act_postprocess = nn.ModuleList([
+            self.act_1_postprocess,
+            self.act_2_postprocess,
+            self.act_3_postprocess,
+            self.act_4_postprocess
+        ])
+
+    def adapt_tokens(self, encoder_tokens):
+        # Adapt tokens
+        x = []
+        x.append(encoder_tokens[:, :])
+        x = torch.cat(x, dim=-1)
+        return x
+
+    def forward(self, encoder_tokens: List[torch.Tensor], image_size):
+            #input_info: Dict):
+        assert self.dim_tokens_enc is not None, 'Need to call init(dim_tokens_enc) function first'
+        H, W = image_size
+        
+        # Number of patches in height and width
+        N_H = H // (self.stride_level * self.P_H)
+        N_W = W // (self.stride_level * self.P_W)
+
+        # Hook decoder onto 4 layers from specified ViT layers
+        layers = [encoder_tokens[hook] for hook in self.hooks]
+
+        # Extract only task-relevant tokens and ignore global tokens.
+        layers = [self.adapt_tokens(l) for l in layers]
+
+        # Reshape tokens to spatial representation
+        layers = [rearrange(l, 'b (nh nw) c -> b c nh nw', nh=N_H, nw=N_W) for l in layers]
+
+        layers = [self.act_postprocess[idx](l) for idx, l in enumerate(layers)]
+        # Project layers to chosen feature dim
+        layers = [self.scratch.layer_rn[idx](l) for idx, l in enumerate(layers)]
+
+        # Fuse layers using refinement stages
+        path_4 = self.scratch.refinenet4(layers[3])
+        path_3 = self.scratch.refinenet3(path_4, layers[2])
+        path_2 = self.scratch.refinenet2(path_3, layers[1])
+        path_1 = self.scratch.refinenet1(path_2, layers[0])
+
+        # Output head
+        out = self.head(path_1)
+        
+        return out

croco/models/head_downstream.py

@@ -0,0 +1,58 @@
+# Copyright (C) 2022-present Naver Corporation. All rights reserved.
+# Licensed under CC BY-NC-SA 4.0 (non-commercial use only).
+
+# --------------------------------------------------------
+# Heads for downstream tasks
+# --------------------------------------------------------
+
+"""
+A head is a module where the __init__ defines only the head hyperparameters.
+A method setup(croconet) takes a CroCoNet and set all layers according to the head and croconet attributes.
+The forward takes the features as well as a dictionary img_info containing the keys 'width' and 'height'
+"""
+
+import torch
+import torch.nn as nn
+from .dpt_block import DPTOutputAdapter
+
+
+class PixelwiseTaskWithDPT(nn.Module):
+    """ DPT module for CroCo.
+    by default, hooks_idx will be equal to:
+    * for encoder-only: 4 equally spread layers
+    * for encoder+decoder: last encoder + 3 equally spread layers of the decoder 
+    """
+
+    def __init__(self, *, hooks_idx=None, layer_dims=[96,192,384,768],
+                 output_width_ratio=1, num_channels=1, postprocess=None, **kwargs):
+        super(PixelwiseTaskWithDPT, self).__init__()
+        self.return_all_blocks = True # backbone needs to return all layers 
+        self.postprocess = postprocess
+        self.output_width_ratio = output_width_ratio
+        self.num_channels = num_channels
+        self.hooks_idx = hooks_idx
+        self.layer_dims = layer_dims
+    
+    def setup(self, croconet):
+        dpt_args = {'output_width_ratio': self.output_width_ratio, 'num_channels': self.num_channels}
+        if self.hooks_idx is None:
+            if hasattr(croconet, 'dec_blocks'): # encoder + decoder 
+                step = {8: 3, 12: 4, 24: 8}[croconet.dec_depth]
+                hooks_idx = [croconet.dec_depth+croconet.enc_depth-1-i*step for i in range(3,-1,-1)]
+            else: # encoder only
+                step = croconet.enc_depth//4
+                hooks_idx = [croconet.enc_depth-1-i*step for i in range(3,-1,-1)]
+            self.hooks_idx = hooks_idx
+            print(f'  PixelwiseTaskWithDPT: automatically setting hook_idxs={self.hooks_idx}')
+        dpt_args['hooks'] = self.hooks_idx
+        dpt_args['layer_dims'] = self.layer_dims
+        self.dpt = DPTOutputAdapter(**dpt_args)
+        dim_tokens = [croconet.enc_embed_dim if hook<croconet.enc_depth else croconet.dec_embed_dim for hook in self.hooks_idx]
+        dpt_init_args = {'dim_tokens_enc': dim_tokens}
+        self.dpt.init(**dpt_init_args)
+
+
+    def forward(self, x, img_info):
+        out = self.dpt(x, image_size=(img_info['height'],img_info['width']))
+        if self.postprocess: out = self.postprocess(out)
+        return out