Spaces:
Running
on
Zero
Running
on
Zero
Commit
•
6284dc0
1
Parent(s):
e7c2780
simplify-inference-logic (#4)
Browse files- Simply inference logic (2bbd3bf560c2d714643ed7eb5fb988409b83cd0a)
Co-authored-by: Chris Maltais <chrismaltais@users.noreply.huggingface.co>
app.py
CHANGED
@@ -1,14 +1,17 @@
|
|
1 |
import os
|
2 |
-
from glob import glob
|
3 |
import cv2
|
4 |
import numpy as np
|
5 |
-
from PIL import Image
|
6 |
import torch
|
7 |
-
from torchvision import transforms
|
8 |
-
from transformers import AutoModelForImageSegmentation
|
9 |
import gradio as gr
|
10 |
import spaces
|
|
|
|
|
|
|
|
|
|
|
11 |
from gradio_imageslider import ImageSlider
|
|
|
|
|
12 |
|
13 |
torch.set_float32_matmul_precision('high')
|
14 |
torch.jit.script = lambda f: f
|
@@ -16,21 +19,21 @@ torch.jit.script = lambda f: f
|
|
16 |
device = "cuda" if torch.cuda.is_available() else "cpu"
|
17 |
|
18 |
|
19 |
-
def array_to_pil_image(image, size=(1024, 1024)):
|
20 |
image = cv2.resize(image, size, interpolation=cv2.INTER_LINEAR)
|
21 |
image = Image.fromarray(image).convert('RGB')
|
22 |
return image
|
23 |
|
24 |
|
25 |
class ImagePreprocessor():
|
26 |
-
def __init__(self, resolution=(1024, 1024)) -> None:
|
27 |
self.transform_image = transforms.Compose([
|
28 |
# transforms.Resize(resolution), # 1. keep consistent with the cv2.resize used in training 2. redundant with that in path_to_image()
|
29 |
transforms.ToTensor(),
|
30 |
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]),
|
31 |
])
|
32 |
|
33 |
-
def proc(self, image):
|
34 |
image = self.transform_image(image)
|
35 |
return image
|
36 |
|
@@ -45,14 +48,17 @@ usage_to_weights_file = {
|
|
45 |
'DIS-TR_TEs': 'BiRefNet-DIS5K-TR_TEs'
|
46 |
}
|
47 |
|
48 |
-
from transformers import AutoModelForImageSegmentation
|
49 |
birefnet = AutoModelForImageSegmentation.from_pretrained('/'.join(('zhengpeng7', usage_to_weights_file['General'])), trust_remote_code=True)
|
50 |
birefnet.to(device)
|
51 |
birefnet.eval()
|
52 |
|
53 |
|
54 |
@spaces.GPU
|
55 |
-
def predict(
|
|
|
|
|
|
|
|
|
56 |
global birefnet
|
57 |
# Load BiRefNet with chosen weights
|
58 |
_weights_file = '/'.join(('zhengpeng7', usage_to_weights_file[weights_file] if weights_file is not None else usage_to_weights_file['General']))
|
@@ -62,32 +68,32 @@ def predict(image, resolution, weights_file):
|
|
62 |
birefnet.eval()
|
63 |
|
64 |
resolution = f"{image.shape[1]}x{image.shape[0]}" if resolution == '' else resolution
|
65 |
-
# Image is a RGB numpy array.
|
66 |
resolution = [int(int(reso)//32*32) for reso in resolution.strip().split('x')]
|
67 |
-
|
68 |
-
|
69 |
-
|
70 |
|
71 |
-
|
72 |
-
|
73 |
-
|
74 |
-
|
75 |
-
images_proc = torch.cat([image_proc.unsqueeze(0) for image_proc in images_proc])
|
76 |
|
|
|
77 |
with torch.no_grad():
|
78 |
-
|
79 |
-
|
80 |
-
|
81 |
-
|
82 |
-
|
83 |
-
|
84 |
-
|
85 |
-
|
86 |
-
|
87 |
-
|
88 |
-
|
89 |
-
|
90 |
-
|
|
|
91 |
|
92 |
|
93 |
examples = [[_] for _ in glob('examples/*')][:]
|
|
|
1 |
import os
|
|
|
2 |
import cv2
|
3 |
import numpy as np
|
|
|
4 |
import torch
|
|
|
|
|
5 |
import gradio as gr
|
6 |
import spaces
|
7 |
+
|
8 |
+
from glob import glob
|
9 |
+
from typing import Optional, Tuple
|
10 |
+
|
11 |
+
from PIL import Image
|
12 |
from gradio_imageslider import ImageSlider
|
13 |
+
from transformers import AutoModelForImageSegmentation
|
14 |
+
from torchvision import transforms
|
15 |
|
16 |
torch.set_float32_matmul_precision('high')
|
17 |
torch.jit.script = lambda f: f
|
|
|
19 |
device = "cuda" if torch.cuda.is_available() else "cpu"
|
20 |
|
21 |
|
22 |
+
def array_to_pil_image(image: np.ndarray, size: Tuple[int, int] = (1024, 1024)) -> Image.Image:
|
23 |
image = cv2.resize(image, size, interpolation=cv2.INTER_LINEAR)
|
24 |
image = Image.fromarray(image).convert('RGB')
|
25 |
return image
|
26 |
|
27 |
|
28 |
class ImagePreprocessor():
|
29 |
+
def __init__(self, resolution: Tuple[int, int] = (1024, 1024)) -> None:
|
30 |
self.transform_image = transforms.Compose([
|
31 |
# transforms.Resize(resolution), # 1. keep consistent with the cv2.resize used in training 2. redundant with that in path_to_image()
|
32 |
transforms.ToTensor(),
|
33 |
transforms.Normalize([0.485, 0.456, 0.406], [0.229, 0.224, 0.225]),
|
34 |
])
|
35 |
|
36 |
+
def proc(self, image: Image.Image) -> torch.Tensor:
|
37 |
image = self.transform_image(image)
|
38 |
return image
|
39 |
|
|
|
48 |
'DIS-TR_TEs': 'BiRefNet-DIS5K-TR_TEs'
|
49 |
}
|
50 |
|
|
|
51 |
birefnet = AutoModelForImageSegmentation.from_pretrained('/'.join(('zhengpeng7', usage_to_weights_file['General'])), trust_remote_code=True)
|
52 |
birefnet.to(device)
|
53 |
birefnet.eval()
|
54 |
|
55 |
|
56 |
@spaces.GPU
|
57 |
+
def predict(
|
58 |
+
image: np.ndarray,
|
59 |
+
resolution: str,
|
60 |
+
weights_file: Optional[str]
|
61 |
+
) -> Tuple[np.ndarray, np.ndarray]:
|
62 |
global birefnet
|
63 |
# Load BiRefNet with chosen weights
|
64 |
_weights_file = '/'.join(('zhengpeng7', usage_to_weights_file[weights_file] if weights_file is not None else usage_to_weights_file['General']))
|
|
|
68 |
birefnet.eval()
|
69 |
|
70 |
resolution = f"{image.shape[1]}x{image.shape[0]}" if resolution == '' else resolution
|
|
|
71 |
resolution = [int(int(reso)//32*32) for reso in resolution.strip().split('x')]
|
72 |
+
|
73 |
+
image_shape = image.shape[:2]
|
74 |
+
image_pil = array_to_pil_image(image, tuple(resolution))
|
75 |
|
76 |
+
# Preprocess the image
|
77 |
+
image_preprocessor = ImagePreprocessor(resolution=tuple(resolution))
|
78 |
+
image_proc = image_preprocessor.proc(image_pil)
|
79 |
+
image_proc = image_proc.unsqueeze(0)
|
|
|
80 |
|
81 |
+
# Perform the prediction
|
82 |
with torch.no_grad():
|
83 |
+
scaled_pred_tensor = birefnet(image_proc.to(device))[-1].sigmoid()
|
84 |
+
|
85 |
+
if device == 'cuda':
|
86 |
+
scaled_pred_tensor = scaled_pred_tensor.cpu()
|
87 |
+
|
88 |
+
# Resize the prediction to match the original image shape
|
89 |
+
pred = torch.nn.functional.interpolate(scaled_pred_tensor, size=image_shape, mode='bilinear', align_corners=True).squeeze().numpy()
|
90 |
+
|
91 |
+
# Apply the prediction mask to the original image
|
92 |
+
image_pil = image_pil.resize(pred.shape[::-1])
|
93 |
+
pred = np.repeat(np.expand_dims(pred, axis=-1), 3, axis=-1)
|
94 |
+
image_pred = (pred * np.array(image_pil)).astype(np.uint8)
|
95 |
+
|
96 |
+
return image, image_pred
|
97 |
|
98 |
|
99 |
examples = [[_] for _ in glob('examples/*')][:]
|