Update app.py

app.py

@@ -4,34 +4,27 @@ from transformers import AutoProcessor, AutoModelForImageClassification
 import gradio as gr
 import pytesseract
 
-def classify_meme(image: Image.Image):
-    # OCR: extract text from image
-    extracted_text = pytesseract.image_to_string(image)
-
-    # Process image with SigLIP2 model
-    inputs = processor(images=image, return_tensors="pt").to(model.device)
-    with torch.no_grad():
-        outputs = model(**inputs)
-        probs = torch.nn.functional.softmax(outputs.logits, dim=-1)
-    predictions = {labels[i]: float(probs[0][i]) for i in range(len(labels))}
-
-    return {
-        "Predictions": predictions,
-        "Extracted Text": extracted_text.strip()
-    }
-
-# Load model and processor from Hugging Face
+# Load model and processor
 model = AutoModelForImageClassification.from_pretrained("google/siglip2-base-patch16-naflex")
 processor = AutoProcessor.from_pretrained("google/siglip2-base-patch16-naflex")
 labels = model.config.id2label
 
+# Classify meme and extract text
 def classify_meme(image: Image.Image):
+    # OCR: extract text from image
+    extracted_text = pytesseract.image_to_string(image)
+
+    # Process image with SigLIP2 model
     inputs = processor(images=image, return_tensors="pt").to(model.device)
     with torch.no_grad():
         outputs = model(**inputs)
         probs = torch.nn.functional.softmax(outputs.logits, dim=-1)
     predictions = {labels[i]: float(probs[0][i]) for i in range(len(labels))}
-    return predictions
+
+    return {
+        "Predictions": predictions,
+        "Extracted Text": extracted_text.strip()
+    }
 
 # Gradio interface
 demo = gr.Interface(
@@ -45,7 +38,5 @@ demo = gr.Interface(
     description="Upload a meme to classify its sentiment and extract text using OCR."
 )
 
-
 if __name__ == "__main__":
-    demo.launch(share = True)
-
+    demo.launch()