Create app.py

app.py

@@ -0,0 +1,100 @@
+import gradio as gr
+import google.generativeai as genai
+from PIL import Image
+
+def extract_malayalam_text(image, api_key):
+    """
+    Extract handwritten Malayalam text from an image using Gemini 2.0 Flash
+    
+    Args:
+        image (PIL.Image): Uploaded image
+        api_key (str): Google AI Studio API Key
+    
+    Returns:
+        str: Extracted text from the image
+    """
+    # Validate inputs
+    if image is None:
+        return "Please upload an image first."
+    
+    if not api_key:
+        return "Please provide a valid API key."
+
+    try:
+        # Configure the Gemini API directly with the provided key
+        genai.configure(api_key=api_key)
+
+        # Use Gemini 2.0 Flash model
+        model = genai.GenerativeModel('gemini-2.0-flash')
+
+        # Generate content from the image
+        response = model.generate_content(
+            [
+                "You are an expert in extracting handwritten Malayalam text from grocery lists. "
+                "Carefully transcribe each item in Malayalam script. "
+                "Provide a clear, accurate list of items. "
+                "If the text is unclear, mention potential uncertainties.", 
+                image
+            ],
+            generation_config=genai.types.GenerationConfig(
+                temperature=0.2,  # Low temperature for precise extraction
+                max_output_tokens=300  # Adjust based on expected list length
+            )
+        )
+
+        # Return the extracted text
+        return response.text
+
+    except Exception as e:
+        return f"An error occurred: {str(e)}"
+
+def create_malayalam_ocr_interface():
+    """
+    Create Gradio interface for Malayalam OCR
+    """
+    with gr.Blocks() as demo:
+        # Title and description
+        gr.Markdown("# Malayalam Handwritten Text Extractor")
+        gr.Markdown("Upload a handwritten Malayalam grocery list image for text extraction.")
+        
+        # API Key input
+        api_key_input = gr.Textbox(
+            label="Google AI Studio API Key", 
+            type="password", 
+            placeholder="Enter your Gemini API key"
+        )
+        
+        # Image upload component
+        image_input = gr.Image(
+            type="pil", 
+            label="Upload Malayalam Grocery List Image"
+        )
+        
+        # Extract button
+        extract_btn = gr.Button("Extract Text")
+        
+        # Output text area
+        output_text = gr.Textbox(
+            label="Extracted Malayalam Text", 
+            lines=10
+        )
+        
+        # Connect the components
+        extract_btn.click(
+            fn=extract_malayalam_text, 
+            inputs=[image_input, api_key_input], 
+            outputs=output_text
+        )
+
+    return demo
+
+def main():
+    # Create and launch the Gradio interface
+    demo = create_malayalam_ocr_interface()
+    demo.launch(
+        share=True,  # Create a public link
+        debug=True   # Show detailed errors
+    )
+
+if __name__ == "__main__":
+    main()