Spaces:
Running
Running
Update app.py
Browse files
app.py
CHANGED
@@ -24,7 +24,7 @@ image_transform = transforms.Compose([
|
|
24 |
|
25 |
def load_image_from_url(url):
|
26 |
try:
|
27 |
-
response = requests.get(url, timeout=10)
|
28 |
response.raise_for_status()
|
29 |
return Image.open(BytesIO(response.content)).convert("RGB")
|
30 |
except Exception:
|
@@ -34,7 +34,7 @@ def load_image_from_url(url):
|
|
34 |
def predict_tags(image_url, threshold=0.5):
|
35 |
image = load_image_from_url(image_url)
|
36 |
if image is None:
|
37 |
-
return
|
38 |
|
39 |
image_tensor = image_transform(image).unsqueeze(0).to(device)
|
40 |
with torch.no_grad():
|
@@ -49,25 +49,36 @@ def predict_tags(image_url, threshold=0.5):
|
|
49 |
results.sort(key=lambda x: x[1], reverse=True)
|
50 |
return results, None
|
51 |
|
52 |
-
def gradio_predict(
|
53 |
-
|
54 |
-
|
55 |
-
|
56 |
-
|
57 |
-
|
58 |
-
|
59 |
-
|
60 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
61 |
|
62 |
demo = gr.Interface(
|
63 |
fn=gradio_predict,
|
64 |
inputs=[
|
65 |
-
gr.Textbox(label="Image URL"),
|
66 |
gr.Slider(0, 1, value=0.5, step=0.01, label="Threshold"),
|
67 |
],
|
68 |
outputs=gr.Textbox(label="Tags"),
|
69 |
-
title="Image Tagging with ViT",
|
70 |
-
description="Paste
|
71 |
)
|
72 |
|
73 |
if __name__ == "__main__":
|
|
|
24 |
|
25 |
def load_image_from_url(url):
|
26 |
try:
|
27 |
+
response = requests.get(url.strip(), timeout=10)
|
28 |
response.raise_for_status()
|
29 |
return Image.open(BytesIO(response.content)).convert("RGB")
|
30 |
except Exception:
|
|
|
34 |
def predict_tags(image_url, threshold=0.5):
|
35 |
image = load_image_from_url(image_url)
|
36 |
if image is None:
|
37 |
+
return None, "Could not load image."
|
38 |
|
39 |
image_tensor = image_transform(image).unsqueeze(0).to(device)
|
40 |
with torch.no_grad():
|
|
|
49 |
results.sort(key=lambda x: x[1], reverse=True)
|
50 |
return results, None
|
51 |
|
52 |
+
def gradio_predict(urls, threshold):
|
53 |
+
url_list = [u.strip() for u in urls.split(",") if u.strip()]
|
54 |
+
output = []
|
55 |
+
|
56 |
+
for url in url_list:
|
57 |
+
tags, error = predict_tags(url, threshold)
|
58 |
+
if error or not tags:
|
59 |
+
output.append({
|
60 |
+
"image_url": url,
|
61 |
+
"error": error or "No tags above threshold."
|
62 |
+
})
|
63 |
+
else:
|
64 |
+
top_tag, top_score = tags[0]
|
65 |
+
output.append({
|
66 |
+
"image_url": url,
|
67 |
+
"tag_name": top_tag,
|
68 |
+
"tag_score": round(top_score, 4)
|
69 |
+
})
|
70 |
+
|
71 |
+
return str(output) # Return as string for textbox display
|
72 |
|
73 |
demo = gr.Interface(
|
74 |
fn=gradio_predict,
|
75 |
inputs=[
|
76 |
+
gr.Textbox(label="Image URL(s) (comma-separated)"),
|
77 |
gr.Slider(0, 1, value=0.5, step=0.01, label="Threshold"),
|
78 |
],
|
79 |
outputs=gr.Textbox(label="Tags"),
|
80 |
+
title="Batch Image Tagging with ViT",
|
81 |
+
description="Paste one or more image URLs separated by commas to get predicted tags using thelabel/240903-image-tagging model.",
|
82 |
)
|
83 |
|
84 |
if __name__ == "__main__":
|