And so on, but the face recognition has three patterns, you really know the difference between them?
First introduce you to a 1:1 as a static comparison, in the financial, huge potential commercial value in the field of information security, such as in the airport security card and id information matching process is badly typical 1:1 scenario, however, the human eye to distinguish rate is only about 95%, and will be affected by external environment, so the airport security personnel through the shift to ensure that the identification accuracy,
Second 1: N is the portrait in massive data in the database to find the current user's face and matching, 1: N with dynamic matching with the characteristics of dynamic contrast refers to the interception based on dynamic video stream to get face data and further process, rather than with sexual identification process is not mandatory and efficiency performance, recognition objects need not to a specific location can do work, because these two features make 1: N authentication model can quickly be born in public security management and VIP customers face recognition scenarios, such as static 1:1, but the difficulty is much higher because the machine is facing overexposed, backlight, side face, long distance challenges, such as
1 N reference:
Finally M, N is for all people in the scene by computer face recognition and portrait ratio on the process of the database, M: N, as a kind of dynamic face, its usage is very high, can be fully applied to a variety of scenarios, such as public security, hospitality, robot applications, etc., but the M: N model there are still a lot of disadvantages, because it must depend on mass face database to run, and due to the large base of identification, equipment factors such as insufficient resolution, make M: N model can produce high error rate would affect recognition results,