Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import gradio as gr
|
2 |
+
from transformers import AutoProcessor, AutoModelForImageClassification
|
3 |
+
import torch
|
4 |
+
from PIL import Image
|
5 |
+
|
6 |
+
model_name = 'e1010101/vit-384-tongue-image'
|
7 |
+
processor = AutoProcessor.from_pretrained(model_name)
|
8 |
+
model = AutoModelForImageClassification.from_pretrained(model_name)
|
9 |
+
|
10 |
+
def classify_image(image):
|
11 |
+
inputs = processor(images=image, return_tensors="pt")
|
12 |
+
with torch.no_grad():
|
13 |
+
outputs = model(**inputs)
|
14 |
+
logits = outputs.logits
|
15 |
+
# Apply sigmoid for multi-label classification
|
16 |
+
probs = torch.sigmoid(logits)[0].numpy()
|
17 |
+
# Get label names
|
18 |
+
labels = model.config.id2label.values()
|
19 |
+
# Create a dictionary of labels and probabilities
|
20 |
+
result = {label: float(prob) for label, prob in zip(labels, probs)}
|
21 |
+
# Sort results by probability
|
22 |
+
result = dict(sorted(result.items(), key=lambda item: item[1], reverse=True))
|
23 |
+
return result
|
24 |
+
|
25 |
+
interface = gr.Interface(
|
26 |
+
fn=classify_image,
|
27 |
+
inputs=gr.inputs.Image(type="pil"),
|
28 |
+
outputs=gr.outputs.Label(num_top_classes=None),
|
29 |
+
title="Multi-Label Image Classification",
|
30 |
+
description="Upload an image to get classification results."
|
31 |
+
)
|
32 |
+
|
33 |
+
|
34 |
+
if __name__ == "__main__":
|
35 |
+
interface.launch()
|