FaceCloud Massive Face Recognition System is specially designed for tens of millions (or even billions) of applications of massive face databases. It uses the advanced RecoFace face recognition core technology developed by Ruiwei, and combines advanced distributed system architecture and large database technology to support real-time face comparison and search for multi-concurrent requests.
Core functions include:
1. Face Registration
It supports batch face registration from database to database. This function can be used to import all face data from public security identity card database at one time.
Form a complete basic face database.
2. Face Search
Search for a specific face in a specified face database by fast comparison (possibly from a photo or a screenshot in a video stream) and return it as required.
The N results with the highest similarity were ranked from high to low.
3. Face comparison
Provides a 1:N face-to-face comparison function based on pictures. Pictures may come from photographs or video screenshots. The result of the comparison is the comparison of the source and each comparison destination.
Similarity value.
4. Face Duplication
Mainly used in ID card duplication, we can find that the same person has more than one ID card or the same person by comparing two or two faces of the ID witness's face.
The ID number corresponds to the anomalies of different faces.
Functional features:
1. Flexible System Expansion
Based on advanced distributed system architecture, support dynamic expansion of face matching server to support tens of millions or even billions of millions of face databases
2. Excellent Face-to-Face Performance
A single face matching server can complete 20 million real-time comparisons per second. It takes less than 200 ms to detect and recognize faces from human faces. On the scale of millions of face databases,
The hit rate of the top 10 of face comparison results is over 92%.
3. High tolerance of face
With the facial posture correction function, the recognition results will not be affected when the face is tilted up and down within 25 degrees.
For face changes, including expression, beard, glasses, hairstyle, age and so on, the algorithm has good adaptability and does not affect the accuracy of recognition.
4. Supporting Mobile Terminal
The system client can run on the mobile terminal based on Android or IOS. It can collect and compare real-time faces through the mobile terminal.