The current front-end video is not all smart cameras, but the cost of large-scale replacement is very high, the technology changes quickly, the hardware smart camera can not keep up with the background software update, and there are a large number of unskilled analysis video.
The surveillance video public security is stored for at least three months, the cell is stored for at least one month, and then a large number of videos are formatted. More than 95% will not be seen, because they simply can’t see it. Chen Xuesong said that he believes that the first step is to solve the problem of video data structuring. Therefore, the strategy of contempt is to assist the government in handling the video of the mass market under the current comprehensive security policy.
For the future front-end intelligence, Chen Xuesong also believes that this is not just a concept of adding intelligence, but a change in camera strategy. The camera that is displayed is different from the camera that is seen to the machine. People pay attention to the average exposure. The scene needs to be balanced, while the machine does not pay attention to the scene. It only focuses on the object, that is, the clarity of the face.