The AI Edge Server is a telecom-grade video surveillance and analysis device. Built on embedded AI technology, it features powerful large-scale video decoding and real-time AI behavior analysis capabilities. It can operate independently offline, bringing intelligent computing to the edge and effectively enhancing the intelligence and response efficiency of monitoring centers.
| Category | Parameter Description |
| Video Input | Supports simultaneous analysis of 8–9 IP camera streams per card, each channel running up to two algorithms; detection accuracy 90%–95% (trainable). Resolution up to 32×32 grid level, range 3–8 m. Compatible with TensorFlow, Caffe, Theano, MXNet, PyTorch (YOLO5). |
| Decoding Resolution | Up to 2 Megapixels |
| Image Capture Storage | Supports up to 4 million captured images |
| Behavior Analysis | Detects and classifies 8–9 concurrent behaviors per video stream |
| Image Transmission | Real-time upload via WebSocket protocol |
| Detection Features | Abnormal event and alarm behavior detection |
| Recognition Zones | Custom region-based detection configuration |
| Smart Park | Facial recognition, whitelist/blacklist management, climbing detection, face capture, e-bike detection in elevators, illegal parking/intrusion detection, fire/smoke recognition, and high-altitude object detection. |
| Smart Campus | Facial recognition, whitelist/blacklist management, climbing and fall detection, face capture, illegal parking/intrusion detection, and fire/smoke recognition. |
| Correctional Facility | Duty attendance monitoring, whitelist/blacklist management, climbing detection, post-duty tracking, personnel fall detection, and regional intrusion alarms. |
| Smart Construction Site | Helmet, reflective vest, uniform, and mask detection; fire/smoke recognition; fire equipment obstruction monitoring. |
| Smart Mining | Belt deviation and foreign object detection, coal flow monitoring, mining vehicle identification, operator absence alert, blockage detection, intrusion alarms, helmet and safety checks, phone/smoking detection, personnel recognition (above/below ground), hoist operation detection, coal size analysis, gas work monitoring, spill and water-level detection, belt scratch identification, “no pedestrian on track” monitoring, camera occlusion, and cable crossing detection. |
| Oil Depot Safety | Fire and smoke recognition, phone use detection, smoking identification, restricted-area intrusion, and fire equipment obstruction monitoring. |
| Network Port | 1× 10M/100M/1000M adaptive Ethernet port |
| Video Output | 2× HDMI |
| Audio Output | 1× Line Out |
| Control Port | 1× RS485 |
| USB Ports / USB | 2× USB 3.0 |
| Power Supply | DC 12V, 1A |
| Power Consumption | <10W (excluding storage) |
| Operating Temperature | -20°C to +80°C |
| Operating Humidity | 5%–95% (non-condensing) |
