app.py

# 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 PIL import Image
import os.path as path
import sys 
import tempfile

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, resize_numpy_image
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
import spaces
from huggingface_hub import hf_hub_download
pl.ion()

HERE_PATH = path.normpath(path.dirname(__file__))  # noqa
sys.path.insert(0, HERE_PATH)  # noqa
# 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 = False
gradio_delete_cache = 7200
print(f'{HERE_PATH}/third_party/ml-depth-pro/checkpoints/')
hf_hub_download(repo_id="apple/DepthPro", filename='depth_pro.pt', local_dir=f'{HERE_PATH}/third_party/ml-depth-pro/checkpoints/')

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)

# @spaces.GPU(duration=180)
def generate_monocular_depth_maps(img_list, depth_prior_name):
    depth_list = []
    focallength_px_list = []
    
    if depth_prior_name=='Depth Pro':
        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().numpy()  # Depth in [m].
          focallength_px=prediction["focallength_px"].cpu()
          depth = cv2.resize(depth[0], image.size, interpolation=cv2.INTER_CUBIC)
          depth_list.append(depth)
          focallength_px_list.append(focallength_px)
          #np.savez_compressed(path_depthpro, depth=depth, focallength_px=prediction["focallength_px"].cpu())  
    elif depth_prior_name=='Depth Anything V2':
        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)
          #print(image.size)
          depth = pipe(image)["predicted_depth"].numpy()
          print(depth.max(),depth.min())
          #depth = cv2.resize(depth[0], image.size, interpolation=cv2.INTER_CUBIC)
          focallength_px = 200
          print(depth.max(),depth.min())
          depth_list.append(depth)
          focallength_px_list.append(focallength_px)
          #np.savez_compressed(path_depthanything, depth=depth)  
    return depth_list, focallength_px_list

@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):
    depth_list, focallength_px_list = generate_monocular_depth_maps(filelist, depth_prior_name)
    imgs = load_images(filelist, depth_list, focallength_px_list, size=image_size, verbose=not silent,traj_format='custom', depth_prior_name=depth_prior_name)
    # pairs = []
    # pairs.append((imgs[0], imgs[1]))
    # pairs.append((imgs[1], imgs[0]))
    scenegraph_type = 'swinstride-5-noncyclic'
    pairs = make_pairs(imgs, scene_graph=scenegraph_type, prefilter=None, symmetrize=True)
    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)  
    output = inference(pairs, model, device, batch_size=batch_size, verbose=not silent)
    mode = GlobalAlignerMode.PointCloudOptimizer  
    scene = global_aligner(output, device=device, mode=mode, verbose=not silent, shared_focal = True, temporal_smoothing_weight=0.01, translation_weight=1.0,
                               flow_loss_weight=0.01, flow_loss_start_epoch=0.1, flow_loss_thre=25, use_self_mask=True,
                               num_total_iter=300, empty_cache= len(filelist) > 72)
    lr = 0.01
    if mode == GlobalAlignerMode.PointCloudOptimizer:
        loss = scene.compute_global_alignment(init='mst', niter=300, schedule='linear', lr=lr)
    # mode = GlobalAlignerMode.PairViewer
    # scene = global_aligner(output, device=device, mode=mode, verbose=not silent)
    
    outfile = get_3D_model_from_scene(tmpdirname, silent, scene, min_conf_thr, as_pointcloud, mask_sky, clean_depth, transparent_cams, cam_size)
      
    return outfile


def run_example(snapshot, min_conf_thr, as_pointcloud, mask_sky, clean_depth, transparent_cams, cam_size, depth_prior_name, inputfiles, **kw):
    return local_get_reconstructed_scene(inputfiles, min_conf_thr, as_pointcloud, mask_sky, clean_depth, transparent_cams, cam_size, depth_prior_name, **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 Align3R</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=0.02, 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=2, minimum=0.0, maximum=20, step=0.01)
        with gradio.Row():
            as_pointcloud = gradio.Checkbox(value=True, label="As pointcloud")
            mask_sky = gradio.Checkbox(value=True, 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=[
                [
                    os.path.join(HERE_PATH, 'example/bear/00000.jpg'),
                    2, True, True, True, False, 0.02, "Depth Anything V2",
                     [os.path.join(HERE_PATH, 'example/bear/00000.jpg'),
                      os.path.join(HERE_PATH, 'example/bear/00001.jpg'),
                      os.path.join(HERE_PATH, 'example/bear/00002.jpg'),
                      ]
                ],
                [
                    os.path.join(HERE_PATH, 'example/breakdance/00000.jpg'),
                    2, True, True, True, False, 0.02, "Depth Anything V2",
                     [os.path.join(HERE_PATH, 'example/breakdance/00000.jpg'),
                      os.path.join(HERE_PATH, 'example/breakdance/00001.jpg'),
                      os.path.join(HERE_PATH, 'example/breakdance/00002.jpg'),
                      os.path.join(HERE_PATH, 'example/breakdance/00003.jpg'),
                      os.path.join(HERE_PATH, 'example/breakdance/00004.jpg'),
                      os.path.join(HERE_PATH, 'example/breakdance/00005.jpg'),
                      os.path.join(HERE_PATH, 'example/breakdance/00006.jpg'),
                      os.path.join(HERE_PATH, 'example/breakdance/00007.jpg'),
                      os.path.join(HERE_PATH, 'example/breakdance/00008.jpg'),
                      os.path.join(HERE_PATH, 'example/breakdance/00009.jpg'),
                      ]
                ],
                [
                    os.path.join(HERE_PATH, 'example/tennis/00000.jpg'),
                    2, True, True, True, False, 0.02, "Depth Anything V2",
                     [os.path.join(HERE_PATH, 'example/tennis/00000.jpg'),
                      os.path.join(HERE_PATH, 'example/tennis/00001.jpg'),
                      os.path.join(HERE_PATH, 'example/tennis/00002.jpg'),
                      os.path.join(HERE_PATH, 'example/tennis/00003.jpg'),
                      os.path.join(HERE_PATH, 'example/tennis/00004.jpg'),
                      os.path.join(HERE_PATH, 'example/tennis/00005.jpg'),
                      os.path.join(HERE_PATH, 'example/tennis/00006.jpg'),
                      os.path.join(HERE_PATH, 'example/tennis/00007.jpg'),
                      os.path.join(HERE_PATH, 'example/tennis/00008.jpg'),
                      os.path.join(HERE_PATH, 'example/tennis/00009.jpg'),
                      ]
                ],
                [
                    os.path.join(HERE_PATH, 'example/camel/00000.jpg'),
                    2, True, True, True, False, 0.02, "Depth Anything V2",
                     [os.path.join(HERE_PATH, 'example/camel/00000.jpg'),
                      os.path.join(HERE_PATH, 'example/camel/00001.jpg'),
                      os.path.join(HERE_PATH, 'example/camel/00002.jpg'),
                      os.path.join(HERE_PATH, 'example/camel/00003.jpg'),
                      os.path.join(HERE_PATH, 'example/camel/00004.jpg'),
                      os.path.join(HERE_PATH, 'example/camel/00005.jpg'),
                      os.path.join(HERE_PATH, 'example/camel/00006.jpg'),
                      os.path.join(HERE_PATH, 'example/camel/00007.jpg'),
                      os.path.join(HERE_PATH, 'example/camel/00008.jpg'),
                      os.path.join(HERE_PATH, 'example/camel/00009.jpg'),
                      ]
                ],
            ],
            inputs=[snapshot, min_conf_thr, as_pointcloud, mask_sky, clean_depth, transparent_cams, cam_size, depth_prior_name, inputfiles],
            outputs=[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)