პროდუქტის აღწერა
• Deep learning algorithm model based on UNV independent intellectual property rights, face recognition accuracy rate > 99%, false rate < 1% • Built-in deep learning dedicated chip, supports local offline recognition, 3,000 face capacity, face whitelist (1∶ N), 10,000 IC card capacity • Fastest recognition time 0.2 seconds, various model merge mode are used to reduce false rate and increase pass rate • WDR, 2MP (1080P) low illumination wide-angle camera for capturing high quality image with various complex lighting scenes • Support anti-spoofing detection based on deep learning algorithm, effective against fraud such as photo and video • Support face metering and human metering for fast adapting to ambient light • Suggested height for face recognition: between 0.8m and 2.2m,face recognition distance: 0.2m to 2m • Support screen sleep mode, keep the minimum brightness to prevent glare at night • Support add up to 6 photos of the base library for a single person • Support video capture, support ONVIF protocol • Support face, card authentication to control door open • Built-in 4G EMMC front end storage, stable and reliable, up to 8,000 events capacity (with images) • Support direct control door lock, exit button, door contact detection to implement access control management • Support tamper protection, support door open timeout and time exceed alarm function to keep door opening during fire alarm active