Edge Deployment
Hardware Requirements
| Component | Minimum | Recommended |
|---|---|---|
| Edge Device | Jetson Nano | Jetson Xavier NX |
| RAM | 4GB | 8GB |
| Storage | 32GB | 64GB |
| Camera Resolution | 720p | 1080p |
| Frame Rate | 15 FPS | 30 FPS |
Supported Cameras
| Camera Type | Compatibility | Notes |
|---|---|---|
| IP Cameras | RTSP streams | Most common |
| USB Cameras | Direct connect | Edge device local |
| Existing CCTV | Via encoder | Retrofit existing |
| PTZ Cameras | Supported | Position presets |
Edge Processing
All vision AI runs on-premise:
| Advantage | Description |
|---|---|
| Low Latency | Under 100ms processing |
| Privacy | Video never leaves premises |
| Offline Capable | Works without internet |
| Bandwidth | No video upload required |
Cloud Sync
Only metadata syncs to cloud:
| Data Type | Synced | Frequency |
|---|---|---|
| Detection Events | Yes | Real-time |
| Aggregated Metrics | Yes | Every 5 min |
| Model Updates | Yes | Daily (download) |
| Video Footage | No | Never |
| Still Images | Optional | On-demand |
Best Practices
- Camera placement - Mount cameras to avoid glare and obstructions
- Lighting - Ensure consistent lighting for better detection
- Model training - Train on YOUR menu items for best accuracy
- Regular calibration - Recalibrate after camera moves
- Privacy notices - Post appropriate signage for customers
- Edge maintenance - Keep edge devices updated and ventilated
Troubleshooting
Low Detection Accuracy
Symptoms: Missed items, false positives
Solutions:
- Check camera focus and cleanliness
- Verify lighting conditions
- Retrain model on recent images
- Adjust confidence thresholds
High Latency
Symptoms: Slow detection, lag in updates
Solutions:
- Check edge device CPU/GPU usage
- Reduce camera resolution if needed
- Verify network connection to edge
- Clear edge device cache