🎥 SmolVLM Realtime Vision

A simple web application for real-time AI vision analysis using SmolVLM-500M-Instruct with live camera feed processing and text-to-speech.

🚀 Quick Start

Prerequisites

• Modern web browser with camera access
• llama-server with SmolVLM running locally

Setup

1. Install and run SmolVLM server:

   # Install llama-server and SmolVLM
   # Run the server
   llama-server -hf ggml-org/SmolVLM-500M-Instruct-GGUF

2. Serve the application:

   # Using Python
   python -m http.server 8000
   
   # Or open index.html directly in your browser

3. Access: Open http://localhost:8000 in your browser

💻 Usage

1. Grant camera permissions when prompted
2. Configure API endpoint (default: http://localhost:8080)
3. Set your instruction (e.g., "What do you see?")
4. Choose request interval (100ms - 2s)
5. Click Start to begin real-time analysis

🎯 Features

• Real-time camera feed processing
• Customizable AI instructions
• Text-to-speech responses
• Adjustable analysis intervals
• Modern, responsive UI

🛡️ Privacy

• Local processing only
• No data storage
• Secure HTTPS connections required
• User-controlled camera permissions

📝 License

MIT License

---

Made with ❤️ for the AI community