Face Recognition 1

Night light supplement supporting photosensitive sensor coordination；
Support serial port, wegen 26, 34 output, output content support configuration;
Adopt dynamic face detection and tracking recognition algorithm which based on video stream
Support equipment to store ten thousand data locally,
cloud platform device support and store 50000 face photo (less than 400 KB), 1 million identification record (0.45 KB), 20 thousands photos which captured at the scene.
LAN device support and store 20 thousands face photo (100 KB), 1 million identification records (including 10 thousands photos which captured at the scene recently) ;
When the face database is 3000 and the false recognition rate is 3/10000, the accuracy of 1:N recognition is 99.7%.；When the face database is 3000, the accuracy rate of 1: N recognition is 99.7% which under the condition of 0.03% misidentification rate.
Fast recognition speed:
Face tracking and detection takes about 20ms time.
Face feature extraction takes about 200ms time.
Face alignment time consuming 0.2ms（1000The average number of people’s library is recognized.），5ms（10000 person library, multiple recognition average value）；

Fast recognition:

it takes about 20ms for face tracking and detection.
it takes about 200ms for face feature extraction.
face matching time: 0.2ms (1000 database, multiple identification and taking the average), 0.5ms (10000 database, multiple identification and taking the average)