China: surveillance system detects people in your Gang

0
10

Who needs face detection? In China, a monitoring system is developed, the citizens at their unique speed detects. The Software should also let intentionally others not to be fooled. In real time that doesn’t work but still.

Article
posted
7. November 2018, At 14:32,
Oliver Nickel

John Cleese will not be able to deceive the monitoring system.
(Image: Pixabay.com/CC0 1.0)

Chinese authorities are working on an image recognition system that recognizes persons of their peculiarities. The technique of Gait Recognition are already used in the public space in Beijing and Shanghai, according to the US news Agency Associated Press (AP). The advantage of this is that individuals cannot be identified, if your face is due to an obstacle or a headpiece is visible. Developed the Software from a company called Watrix.

The job market

  1. about dueren Hoff GmbH, Munich
  2. Abel Mobilfunk GmbH & co. KG, mount angel

Its CEO Huang Yongzhen does not respond to his product: “You need the cooperation of people to find identity,” he said in an Interview with AP. “Gait can not be easily deceived, so by Limping, with splayed feet or humps, since we analyze the properties of the entire body.” The adoption of fingerprints that each Person has a unique gait, as well as clear finger.

More computing power needed

In order to detect a walking pattern, however, it is necessary to analyze more than just an image. A video sequence requires more computational power than, for example, facial recognition programs, as they are also used in China and other countries. For this reason, the Software is not working yet in real time. Instead, users of video material load in the System. An analysis of the movement pattern is continuing, according to AP, about ten minutes for one hour of source material. For recording, conventional security cameras can be used.

Although the technique is not new, it seems in China has a strong demand for AI-systems for population monitoring. Originally, the Japanese, and the US have developed, the British and Americans recognition software for the analysis of the running properties. The University of Osaka has been testing since 2013 a practical use.