Update app.py
Browse files
app.py
CHANGED
@@ -1,75 +1,60 @@
|
|
1 |
-
import
|
2 |
import torch
|
3 |
-
|
4 |
from PIL import Image
|
5 |
-
import
|
6 |
|
7 |
-
#
|
8 |
-
device =
|
9 |
-
model_path = "model/cloth_segm.pth"
|
10 |
|
11 |
-
|
12 |
-
|
13 |
-
|
14 |
-
|
15 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
16 |
|
17 |
-
def process_image(
|
18 |
"""Process input image and return segmentation mask"""
|
19 |
-
if
|
20 |
-
raise gr.Error("Please upload
|
21 |
|
22 |
try:
|
23 |
-
|
24 |
-
if not isinstance(input_img, Image.Image):
|
25 |
-
input_img = Image.fromarray(input_img)
|
26 |
-
|
27 |
-
# Generate mask
|
28 |
-
output_mask = generate_mask(input_img, net=net, palette=palette, device=device)
|
29 |
-
return output_mask
|
30 |
except Exception as e:
|
31 |
-
raise gr.Error(f"
|
32 |
|
33 |
-
#
|
34 |
-
|
35 |
-
|
36 |
-
|
37 |
-
|
38 |
-
|
|
|
|
|
|
|
39 |
|
40 |
with gr.Row():
|
41 |
with gr.Column():
|
42 |
-
input_image = gr.Image(
|
43 |
-
sources=["upload", "webcam"],
|
44 |
-
type="pil",
|
45 |
-
label="Input Image",
|
46 |
-
interactive=True
|
47 |
-
)
|
48 |
submit_btn = gr.Button("Process", variant="primary")
|
49 |
-
|
50 |
with gr.Column():
|
51 |
-
output_image = gr.Image(
|
52 |
-
label="Segmentation Result",
|
53 |
-
interactive=False
|
54 |
-
)
|
55 |
-
|
56 |
-
# Examples section (optional)
|
57 |
-
example_dir = "input"
|
58 |
-
if os.path.exists(example_dir):
|
59 |
-
example_images = [
|
60 |
-
os.path.join(example_dir, f)
|
61 |
-
for f in os.listdir(example_dir)
|
62 |
-
if f.lower().endswith(('.png', '.jpg', '.jpeg'))
|
63 |
-
]
|
64 |
-
|
65 |
-
gr.Examples(
|
66 |
-
examples=example_images,
|
67 |
-
inputs=[input_image],
|
68 |
-
outputs=[output_image],
|
69 |
-
fn=process_image,
|
70 |
-
cache_examples=True,
|
71 |
-
label="Example Images"
|
72 |
-
)
|
73 |
|
74 |
submit_btn.click(
|
75 |
fn=process_image,
|
@@ -77,10 +62,5 @@ with gr.Blocks(title="Cloth Segmentation") as demo:
|
|
77 |
outputs=output_image
|
78 |
)
|
79 |
|
80 |
-
# Launch with appropriate settings
|
81 |
if __name__ == "__main__":
|
82 |
-
demo.launch(
|
83 |
-
server_name="0.0.0.0",
|
84 |
-
server_port=7860,
|
85 |
-
show_error=True
|
86 |
-
)
|
|
|
1 |
+
import os
|
2 |
import torch
|
3 |
+
import gradio as gr
|
4 |
from PIL import Image
|
5 |
+
from process import load_seg_model, get_palette, generate_mask
|
6 |
|
7 |
+
# Device selection
|
8 |
+
device = 'cuda' if torch.cuda.is_available() else 'cpu'
|
|
|
9 |
|
10 |
+
def load_model():
|
11 |
+
"""Load model with Hugging Face Spaces compatible paths"""
|
12 |
+
model_dir = 'model'
|
13 |
+
checkpoint_path = os.path.join(model_dir, 'cloth_segm.pth')
|
14 |
+
|
15 |
+
# Verify model exists (must be pre-uploaded to HF Spaces)
|
16 |
+
if not os.path.exists(checkpoint_path):
|
17 |
+
raise FileNotFoundError(
|
18 |
+
f"Model not found at {checkpoint_path}. "
|
19 |
+
"Please upload the model file to your Space's repository."
|
20 |
+
)
|
21 |
+
|
22 |
+
try:
|
23 |
+
net = load_seg_model(checkpoint_path, device=device)
|
24 |
+
palette = get_palette(4)
|
25 |
+
return net, palette
|
26 |
+
except Exception as e:
|
27 |
+
raise RuntimeError(f"Model loading failed: {str(e)}")
|
28 |
+
|
29 |
+
# Initialize model (will fail fast if there's an issue)
|
30 |
+
net, palette = load_model()
|
31 |
|
32 |
+
def process_image(img: Image.Image) -> Image.Image:
|
33 |
"""Process input image and return segmentation mask"""
|
34 |
+
if img is None:
|
35 |
+
raise gr.Error("Please upload an image first")
|
36 |
|
37 |
try:
|
38 |
+
return generate_mask(img, net=net, palette=palette, device=device)
|
|
|
|
|
|
|
|
|
|
|
|
|
39 |
except Exception as e:
|
40 |
+
raise gr.Error(f"Processing failed: {str(e)}")
|
41 |
|
42 |
+
# Gradio interface
|
43 |
+
title = "Cloth Segmentation Demo"
|
44 |
+
description = """
|
45 |
+
Upload an image to get cloth segmentation using U2NET.
|
46 |
+
"""
|
47 |
+
|
48 |
+
with gr.Blocks() as demo:
|
49 |
+
gr.Markdown(f"## {title}")
|
50 |
+
gr.Markdown(description)
|
51 |
|
52 |
with gr.Row():
|
53 |
with gr.Column():
|
54 |
+
input_image = gr.Image(type="pil", label="Input Image")
|
|
|
|
|
|
|
|
|
|
|
55 |
submit_btn = gr.Button("Process", variant="primary")
|
|
|
56 |
with gr.Column():
|
57 |
+
output_image = gr.Image(type="pil", label="Segmentation Result")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
58 |
|
59 |
submit_btn.click(
|
60 |
fn=process_image,
|
|
|
62 |
outputs=output_image
|
63 |
)
|
64 |
|
|
|
65 |
if __name__ == "__main__":
|
66 |
+
demo.launch()
|
|
|
|
|
|
|
|