Real-time Visual Description Service for Xreal AR Glasses
LiveLens provides real-time visual descriptions of your camera feed using Vision Language Models (VLM). Designed for Xreal AR glasses with smartphone camera streaming.
- Real-time Camera Streaming: Access smartphone camera via WebRTC
- Dual AI Mode:
- 🔷 On-device VLM (MediaPipe/TensorFlow.js)
- ☁️ Cloud VLM (Gemini API)
- Simple Web App: No installation required, runs in browser
- Responsive UI: Works on smartphones and AR displays
- Clone this repository
git clone https://github.com/wwlapaki310/LiveLens.git
cd LiveLens- Open
index.htmlin your browser (HTTPS required for camera access)
# Using Python's HTTP server
python3 -m http.server 8000- For Gemini API mode, add your API key in
config.js
- Modern web browser (Chrome, Safari, Edge)
- HTTPS connection (for camera access)
- Gemini API key (for cloud mode)
See docs/architecture.md for detailed system design.
┌─────────────┐
│ Xreal AR │ ← Display descriptions
└──────┬──────┘
│
┌──────▼──────┐
│ Smartphone │
│ Browser │
│ ┌─────────┐ │
│ │ Camera │ │
│ └────┬────┘ │
│ │ │
│ ┌────▼────┐ │
│ │LiveLens │ │
│ │Web App │ │
│ └────┬────┘ │
└──────┼──────┘
│
┌───▼────┐
│ VLM │
│ Engine │
└────────┘
On-device / Cloud
- Select AI Mode: Choose between on-device or cloud processing
- Start Camera: Grant camera permissions
- View Descriptions: Real-time visual descriptions appear on screen
- Adjust Settings: Configure capture interval and description detail
- Frontend: Vanilla JavaScript (ES6+)
- Camera: WebRTC getUserMedia API
- On-device AI: MediaPipe Vision Tasks
- Cloud AI: Google Gemini API
- Styling: CSS3 with Flexbox
MIT License - See LICENSE file for details
Developed by Satoru Wada (Sony Semiconductor Solutions)