If the wanted person's face matches the surveillance camera's screen, the technology can immediately record his or her clothing and build images from the rear and obliquely at an angle. This technique can be used to track multiple cameras at a commercial facility or the like. , plan to put into practice in 2018.
This technique first looks for a person who is consistent with a pre-registered face photo from the surveillance camera image being shot, and then memorizes the features of the clothing and body according to the screen after finding it.
If the surveillance camera can take a frontal view, it can learn the image from the side and rear through AI's deep learning. This technology can find a specific person from the pictures captured from all directions. Even if the camera only takes one picture , also can instantly determine if it is not the person to be traced.