Spaces:
Running
Running
| import gradio as gr | |
| import os | |
| from openai import OpenAI | |
| import re | |
| def extract_medicine_names(api_key, image_path, model_choice): | |
| """ | |
| Extract medicine names from a prescription image using OpenRouter API | |
| """ | |
| if not api_key or api_key.strip() == "": | |
| return "Please provide a valid OpenRouter API key." | |
| if not image_path: | |
| return "Please upload an image." | |
| try: | |
| # Initialize the OpenAI client with OpenRouter base URL | |
| client = OpenAI( | |
| base_url="https://openrouter.ai/api/v1", | |
| api_key=api_key, | |
| ) | |
| # Select the model based on user's choice | |
| if model_choice == "Llama 4 Maverick": | |
| model = "meta-llama/llama-4-maverick:free" | |
| else: # Default to Kimi VL | |
| model = "moonshotai/kimi-vl-a3b-thinking:free" | |
| # Prepare image for API | |
| with open(image_path, "rb") as image_file: | |
| import base64 | |
| image_data = base64.b64encode(image_file.read()).decode("utf-8") | |
| # Create the completion | |
| completion = client.chat.completions.create( | |
| extra_headers={ | |
| "HTTP-Referer": "https://medicine-extractor-app.com", | |
| "X-Title": "Medicine Name Extractor", | |
| }, | |
| model=model, | |
| messages=[ | |
| { | |
| "role": "system", | |
| "content": "You are a specialized medical assistant. Your task is to analyze prescription images and extract ONLY the names of medicines/medications. Return them as a clear, numbered list without any other commentary. If you cannot identify any medicine names, state that clearly." | |
| }, | |
| { | |
| "role": "user", | |
| "content": [ | |
| { | |
| "type": "text", | |
| "text": "Please extract and list ONLY the names of medicines or medications from this prescription image." | |
| }, | |
| { | |
| "type": "image_url", | |
| "image_url": { | |
| "url": f"data:image/jpeg;base64,{image_data}" | |
| } | |
| } | |
| ] | |
| } | |
| ] | |
| ) | |
| # Get the response | |
| result = completion.choices[0].message.content | |
| # Process to ensure we're only returning medicine names | |
| # This is a basic processing step - the system prompt should already help focus the response | |
| if "no medicine" in result.lower() or "cannot identify" in result.lower(): | |
| return "No medicine names were identified in the prescription image." | |
| return result | |
| except Exception as e: | |
| return f"Error: {str(e)}" | |
| # Define the Gradio interface | |
| with gr.Blocks(title="Medicine Name Extractor", theme=gr.themes.Soft()) as app: | |
| gr.Markdown("# Medicine Name Extractor") | |
| gr.Markdown("Upload a prescription image and the app will extract medication names using AI vision models.") | |
| with gr.Row(): | |
| with gr.Column(): | |
| api_key = gr.Textbox( | |
| label="OpenRouter API Key", | |
| placeholder="Enter your OpenRouter API key here", | |
| type="password" | |
| ) | |
| model_choice = gr.Radio( | |
| ["Llama 4 Maverick", "Kimi VL"], | |
| label="Select AI Model", | |
| value="Kimi VL" | |
| ) | |
| image_input = gr.Image( | |
| label="Upload Prescription Image", | |
| type="filepath" | |
| ) | |
| submit_btn = gr.Button("Extract Medicine Names", variant="primary") | |
| with gr.Column(): | |
| output = gr.Textbox( | |
| label="Extracted Medicine Names", | |
| lines=10 | |
| ) | |
| # Handle the submission | |
| submit_btn.click( | |
| fn=extract_medicine_names, | |
| inputs=[api_key, image_input, model_choice], | |
| outputs=output | |
| ) | |
| gr.Markdown(""" | |
| ## How to use | |
| 1. Enter your OpenRouter API key | |
| 2. Select an AI vision model | |
| 3. Upload a clear image of a medical prescription | |
| 4. Click "Extract Medicine Names" | |
| ## Privacy Note | |
| Your prescription images are processed securely. No data is stored on our servers. | |
| """) | |
| # Launch the app | |
| if __name__ == "__main__": | |
| app.launch() |