ZKTeco BioServer 2
The BioServer 2 mini-server can be used as a back-end server for facial recognition devices under small scale application. BioServer 2 receives video streams from front-end cameras or captured photo streams, then perform 1: N face comparison, and reports or synchronizes the results to access control, time & attendance or surveillance system for further usage. With strong performance and directional optimization of the server-level facial recognition algorithm, the speed of facial recognition can easily reach 50 faces per second.
The BioServer 2 mini-server is equipped with a tensor processor which provides up to 3.2 TFLOPS dual precision peak computing ability and embedded 16MB of SRAM, while greatly improves the reasoning performance of artificial intelligence, it also reduces the power demand. The processor chip has reached millisecond-grade response in mobile lightweight universal face benchmark model Mobile-net latency test which is only a half or a quarter of industry mainstream chip.
Hardware
| Dimensions (W*D*H) | 115*210*45mm |
| Network interface | 1* gigabit ethernet port |
| Buttons |
Reset button; press and hold the button for 5s to restart the device; press and hold for 15s to reset all parameters to initial factory values |
| PWM | Supported |
| RAM | 8GB |
| Flash | 32GB |
| Average / peak power | 50W / 80W |
| Cooling system | Spontaneous cooling |
| Power | DC 12V; AC 100-240V |
| Storage | Local and cloud storage |
System Parameters
| Operating system | Linux |
| Maximum storage | 30,000 facial templates |
| Recognition speed | 50 pictures / second |
| Video processing capability | 16 channels |
| Video input standard | H.264 |
| Cache capacity for facial picture | 8GB |
| Result’s return time for real-time facial capturing | 32GB |
| Result’s return time for real-time facial comparison | 50W / 80W |
| Algorithm Parameters | |
| Maximum number of detected faces per frame | 20 |
| Maximum number of recognized faces per frame | 8 |
| Pose angle tolerance for face detection | ±60° (yaw, pitch, roll) |
| Pose angle tolerance for facial recognition | ±45° (yaw, pitch, roll) |
| Minimum face size for capturing | 30 * 30 pixel |
| Minimum face size for recognition | 70 * 70 pixel |
| Dynamic facial expression recognition | Available |
| Facial recognition accuracy rate | 98.5% |
Environmental Requirements
| Applicable location | Indoor |
| Storage temperature | -40 ~ 70 °C |
| Working temperature | 0 ~ 45 °C |
| Working humidity |
10% ~ 90% (non-condensing) |