Zhcloudlink
Products
AI Camera
A black-light, ultra-wide-dynamic-range, high-density face-recognition camera, with 2T compute, face recognition, full-color night vision, and GB28181.
AI camera
Model: AIC-06 · Black-light ultra-wide dynamic range high-density face recognition
A black-light, ultra-wide-dynamic-range, high-density face-recognition camera, with 2T compute, face recognition, full-color night vision, and GB28181.
1. Product features
- Built-in HiSilicon high-performance DV500 chip, large-format high-sensitivity 1/1.8” CMOS
- Resolution 2688*1520@25fps
- H264/H.265 efficient compression, saving bandwidth and storage, low bitrate, low latency
- Infrared, full-color, and smart full-color night vision, with soft/hard photosensor switching
- 3D digital noise reduction, WDR/HDR, and backlight compensation.
- Built-in MIC and built-in amplifier. An external speaker can play alarm voice. The content can be customized by scene, with live-person recording and text-to-Mandarin
- 2T compute, face and person detection, face recognition, and capture on interval, leave, and enter
- ONVIF access, GB28181-2016, GAT-1400, P2P, and similar protocols
- Change the password at first boot to keep it secure
- TF card storage: up to 256G (GB remote playback supported)
- External connection supported, 2 alarm outputs / 2 outputs / 1 external audio channel
- Optional 4G wireless transmission.
2. Product specifications:
| Product name | Black-light ultra-wide-dynamic-range high-density face camera |
|---|---|
| System architecture | Embedded Linux design, Hi3516 DV500 dual-core Cortex-A55@850MHz + external 16Gb DDR 2T compute, hardware compression, watchdog, FLASH: 2Gb Nand |
| Image sensor | Large-format 1/1.8”, black-light high-sensitivity CMOS |
| Lens | Large-aperture F1.0 6mm (default), optional (4mm,8mm,12mm) |
| Day/night mode | ICR switch, auto/color/mono/external trigger |
| Wide dynamic range | Wide dynamic range supported; HDR supported; (range: 140 dB) |
| 3D noise reduction | Off/low/medium/higher/ |
| Mirror | Off/horizontal mirror/vertical mirror/180° rotation |
| Audio and video encoding | |
| Video codec | H.265/H.264 |
| Encoding | Main stream: 2688*1520@25fps/ 2560*1440@25fps / 1920*1080@25fps Sub stream: 1280*720@15fps / 960*540@15fps / 640*360@25fps |
| Video bitrate | Constant bitrate, variable bitrate (192kbps~8000kbps) |
| Video type | Combined stream/video stream |
| Audio codec | G.711A / AAC 64Kpbs |
| Face detection | |
| Face detection count | Up to ≤60 face targets at once |
| Face detection size | Face size ≥40 pixels |
| Snapshot response | ≤1s |
| Face snapshot deduplication | After a face leaves, the best face snapshot is selected automatically to reduce duplicate captures |
| Face scene snapshot | Scene snapshots can be captured and uploaded |
| Face tracking | Face detection and tracking supported |
| Trigger | Video trigger |
| Face comparison | |
| Comparison type | Liveness comparison; printed paper and phone photo/video spoofs are filtered automatically |
| Face library capacity | Built-in ≤30000 faces; blacklist and whitelist watchlists supported |
| Face comparison recognition rate | 10000-person library: recognition rate >99% 30000-person library: recognition rate >98 |
| Face comparison false-accept rate | 10000-person library: false-accept rate <0.2% 30000-person library: false-accept rate <0.5% |
| Response time | <=200ms |
| Face pose | Face pose recognition supported (look up, look down, turn left, turn right) |
| Expression | Slight expressions do not affect face comparison |
| Detection type | Liveness detection; printed paper and phone photo/video spoofs are filtered automatically |
| Face library capacity | Built-in ≤30000 faces; blacklist and whitelist watchlists supported |
| Network functions | |
| Network protocols | TCP/IP, HTTP, DHCP, RTSP, NTP |
| Interface protocols | ONVIF, GB28181/2016, GAT-1400, P2P |
| Simultaneous preview streams | 5 |
| Security mode | Authorized username and password, RTSP authentication |
| Motion detection | Set detection regions, with multiple sensitivity levels |
| Event linkage | Snapshot, audio linkage, alarm-light linkage |
| On-screen text (OSD) | Title, time, and date overlay. OSD supports 7 custom OSD groups |
| System upgrade | Tool upgrade and online OTA upgrade |
| Interfaces | |
| Network interface | RJ45 10M/100M adaptive Ethernet port |
| Audio | Built-in MIC interface, built-in amplifier interface + 1 external speaker interface |
| Power interface | 1 × DC 12V power interface |
| Alarm interface | 2 alarm inputs/2 alarm outputs/ |
| Reset | Hardware reset supported |
| General specifications | |
| Operating temperature / humidity | -40℃~+80℃ / ≤95% (no condensation) |
| Operating voltage | DC 12V ; |
| Maximum power | MAX: 6W |
| Module size | 186.00mm*97.76mm |
3. Application functions
4. Algorithm customization
Custom development flow for deep-learning AI vision algorithms: (algorithms can be deployed at the edge or in the cloud. Edge and cloud as one system, with edge-cloud coordination.)
1. Annotate images and video sampled on site;
5. Application scenes
(Deployment scenes need to be implemented together with a backend platform)
Smart community / campus
Algorithms and product use: parking in a fire lane, area intrusion, fall detection, smoke and open-flame detection, line crossing, roadside vending, person counting, e-bike-in-elevator detection, and similar algorithms. Widely used for identity checks of people and vehicles entering and leaving, illegal parking, environmental sanitation checks, other violations and uncivil behavior, and fire-hazard detection.
Algorithms and product use: smoke and fire detection, hazardous-area detection, loitering detection, person detection, person lingering, illegal fishing / stranger intrusion, fall detection, license-plate recognition, and similar algorithms. Widely used for villas, courtyards, fish ponds, farms, planting and breeding bases, and unattended stores.
Food-service stores
Algorithms and product use: passenger-flow statistics, VIP customer recognition, smoking detection, phone-call detection, rodent detection, chef uniform and hat detection, open flame with nobody present, operating-procedure checks, smoke and open-flame detection, overflowing trash, and similar algorithms. Widely used in schools, food service, hospitals, and similar sites.
Smart campus
Student behavior analysis
Jump detection, running detection, fall detection, running detection, climbing detection, fighting detection, smoke and fire detection, smoking detection, phone detection, line-accompanying detection, area detection;
Classroom behavior analysis
Hand raise, stand up, head on desk, reading, writing, look up, look down, using a phone
Teacher behavior analysis
Board writing, stepping on and off the platform, using a phone, making a call, whether seated, presenting a PPT
Questions
What kind of camera is the AIC-06?
The AIC-06 is a black-light, ultra-wide-dynamic-range, high-density face-recognition camera. The image sensor is a 1/1.8-inch black-light CMOS. The default lens is a large-aperture F1.0 6mm, with optional 4mm, 8mm, or 12mm. The main stream supports 2688×1520 at 25fps.
What are the AIC-06 chip and compute?
The main controller is the Hi3516 DV500, a dual-core Cortex-A55 at 850MHz, with 2T compute. The wide-dynamic-range span is 140dB, and HDR is supported.
How large is the AIC-06 face library?
It can be configured with no fewer than 30000 faces, and it supports blacklists and whitelists. The recognition rate is greater than 99% on a 10000-person library and greater than 98% on a 30000-person library, with a response time of no more than 200ms. Liveness detection can filter printed photos and video spoofs.
What are the AIC-06 night vision and network protocols?
It supports infrared, full-color, and smart full-color night vision. Interface protocols include ONVIF, GB28181-2016, GAT-1400, and P2P. A TF card is supported up to 256G.
What are the AIC-06 power supply and dimensions?
Power is DC 12V, and maximum consumption is about 6W. The module measures 186.00mm × 97.76mm. Operating temperature is -40℃ to +80℃, and humidity does not exceed 95% without icing.