Trends in Intelligent Security: Face Recognition + Feature Recognition
The curren security Camera Monitoring system can structure the pedestrians and their attributes in the picture so that the machine can recognize some basic attributes such as sex, age, color of clothes, type of clothes, etc., if the video image is captured from the target side, Can identify some specific features of the face, such as whether to wear a mask, glasses, beard type and so on.
But this kind of information collection and face recognition is different, Guling deep pupil CEO Zhao Yong said it was a short-term, the resolution is not high when a way to get information. Coincidentally, there are many experts in foreign countries today, the vast majority of ordinary security cameras can not meet the resolution of face recognition conditions, it seems that feature recognition than the face recognition in the capture of criminals to be more convenient.
Feature recognition allows the camera to continue to track the corresponding person (the system is set to the object of interest) if the camera can not capture the face, according to foreign media, this method makes it easy to catch some terrorist attacks. Of course, there is a big premise here is the attacker in the attack after maintaining the same clothes.
So feature recognition can really replace face recognition? There are two features that must be emphasized here. First, feature recognition provides more "false" intelligence than face recognition, and second, when the characteristics of a person change, their work efficiency will be reduced. So feature recognition is more suitable for immediate tracking of short-term situations in the scene, and not for long-term tracking.
More current monitoring vendors as an important complement to face recognition can help people in a wide range of camera networks to track a person, dig out more suspects characteristics. According to foreign media, although the current feature recognition still loopholes, but with the development of artificial intelligence and depth of learning, the future of intelligent identification will become the standard smart camera. With the introduction of many foreign R & D institutions have begun to start gait recognition, and through the 3D depth sensor to improve the accuracy of gait recognition. Based on this, the author also has reason to believe that the future feature recognition and will not lag behind the face to identify so much, but also to more police officers reconnaissance means more scientific and technological.
From Security knowledge network 2017-03-08