Face Recognition
Face (Facial Recognition), is to get the facial image of the user by the video capture device, and then use an algorithm to calculate the core features of its analysis of the position of the face, and the face angle, and thus has its own database of templates for comparison Yeah, that determine the true identity of the user. One training face recognition face recognition regions based on local features. The first step, the need for the local region is defined; a second step of extracting features of the face partial region, based on a transform matrix obtained after the training sample face image vector mapping human face feature vector; a third step, the local feature selection (optional); after step is to classify. Use more classifiers in combination classifiers, each corresponding to a local feature classifier can be used after a weighted voting or other linear manner to obtain the final recognition result. Face Recognition integrated use of digital image / video processing, pattern recognition, computer vision and other technologies, the core technology is the face recognition algorithm.
principle
Principle of face identification: system input, or generally a series of face images contained undetermined status, and the face image or face database corresponding encoded in several known identity, and its output is a series of similarity score, indicating the identity of the person to be recognized face.
Offline Recognition
Deployment recognition algorithm has two modes: one is the terminal apparatus, the other is a local server.
Equipment deployed in terminal mode: stand-alone operation, face entered in equipment operation, face database stored in the storage device, face database size depends on the size of the device memory chip capacity, generally no more than ten thousand. Face recognition than the complete results directly on the device. Applications include: attendance and access control
Deployed in local server mode: internal linking LAN operation, the appliance to crawl data back to the local server for comparison, then returned to the device than the results. This mode broad range of applications, but also more popular. Advantage is that the server computing ability, large face database, fast identification speed, back-end systems management and control, to facilitate data management staff.
Applications include: attendance and access control; conference room; pedestrian access gates; face pay; staff behavior analysis and so on.
Screening than from self face database, the entire process to complete the operation than in the local server, without passing outside the network, the real closed-loop management of self-built face database. As for the issues of great concern to the user, such as a human face than use the network can not be broken, face than storage capacity is limited, can not be integrated with other business systems, you can not worry about. As the face verification system algorithm deployed locally, there is no net net can be used; in addition, the system is suitable for applications of all sizes, and can be made interactive interface with any business system. Not only allows accurate verification of identity management, it can also extract data from consumer behavior analysis to provide basis for decision making more humane management.
Offline Device
Cloud Online Face Recognition
Gripping the front end of the device data, in comparison to identify cloud services, and then returns the result to the terminal than the device. After the end of APP, PC clients, web side, end, etc. can WeChat H5 identification request sent by the HTTP web service interface call or identification of the service, and upload the face recognition is completed on the server side comparison, comparing the results returned witnesses. The biggest difference between the cloud and the offline version of the algorithm is that the cloud version of the need to connect to external networks, because face recognition, face database is deployed on cloud Ali, Baidu cloud substandard web server.
Applications include: tourist scenic park; large-scale activities on-site; Station; the public domain and so on.