Update app.py

app.py

@@ -13,47 +13,55 @@ model = MllamaForConditionalGeneration.from_pretrained(
 processor = AutoProcessor.from_pretrained(ckpt)
 
 @spaces.GPU
-def extract_text(image):
-    # Convert image to RGB
-    image = Image.open(image).convert("RGB")
-    
-    # Create message structure
-    messages = [
-        {
-            "role": "user",
-            "content": [
-                {"type": "text", "text": "Extract handwritten text from the image and output only the extracted text without any additional description or commentary in output"},
-                {"type": "image"}
-            ]
-        }
-    ]
-    
-    # Process input
-    texts = processor.apply_chat_template(messages, add_generation_prompt=True)
-    inputs = processor(text=texts, images=[image], return_tensors="pt").to("cuda")
-
-    
-    # Generate output
-    outputs = model.generate(**inputs, max_new_tokens=250)
-    result = processor.decode(outputs[0], skip_special_tokens=True)
-
-    # Clean up the output to remove the prompt and assistant text
-    if "assistant" in result.lower():
-        result = result[result.lower().find("assistant") + len("assistant"):].strip()
-    
-    # Remove any remaining conversation markers
-    result = result.replace("user", "").replace("Extract handwritten text from the image and output only the extracted text without any additional description or commentary in output", "").strip()
-    
-    return result
+def extract_text(image_path):
+    try:
+        # Handle different types of image inputs
+        if isinstance(image_path, dict) and "path" in image_path:
+            image_path = image_path["path"]
+        
+        # Convert image to RGB
+        image = Image.open(image_path).convert("RGB")
+        
+        # Create message structure
+        messages = [
+            {
+                "role": "user",
+                "content": [
+                    {"type": "text", "text": "Extract handwritten text from the image and output only the extracted text without any additional description or commentary in output"},
+                    {"type": "image"}
+                ]
+            }
+        ]
+        
+        # Process input
+        texts = processor.apply_chat_template(messages, add_generation_prompt=True)
+        inputs = processor(text=texts, images=[image], return_tensors="pt").to("cuda")
+        
+        # Generate output
+        outputs = model.generate(**inputs, max_new_tokens=250)
+        result = processor.decode(outputs[0], skip_special_tokens=True)
+        
+        # Clean up the output
+        if "assistant" in result.lower():
+            result = result[result.lower().find("assistant") + len("assistant"):].strip()
+        
+        result = result.replace("user", "").replace("Extract handwritten text from the image and output only the extracted text without any additional description or commentary in output", "").strip()
+        
+        return f"\n{result}\n"
+        
+    except Exception as e:
+        return f"Error processing image: {str(e)}"
 
 # Create Gradio interface
 demo = gr.Interface(
     fn=extract_text,
     inputs=gr.Image(type="filepath", label="Upload Image"),
-    outputs=gr.Textbox(label="Extracted Text"),
+    outputs=gr.Textbox(label="Extracted Text", lines=10),  # Increased lines for better visibility
     title="Handwritten Text Extractor",
     description="Upload an image containing handwritten text to extract its content.",
+    examples=[],  # Add example images here if needed
 )
 
-# Launch the app
+# Launch the app with queue enabled
+demo.queue()  # Enable queuing
 demo.launch(debug=True)