AI CCTV DETECTING DANGER AND CRIME SCENES

[Anchor Lead]

Most crimes with wide social repercussions show preceding signs. Now scientists are developing an artificial intelligence technology that enables CCTVs to detect such danger signs and predict crimes. This new technology id driving the evolution of CCTVs from a simple surveillance tool to a crime deterrence.

[Pkg]

A man follows a woman, keeping a few meters of distance between them. When she enters a building, he suddenly attacks her from the back. The police arrested an active police officer as a prime suspect some 20 days after the incident. But, the future seems more promising. A research team is developing a technology that can detect warning signs of such dangers to successfully prevent the crime from happening in the first place. The Electronics and Telecommunications Research Institute is developing an artificial intelligence CCTV that is capable of tracking a potential suspect when it detects unusual signs such as persistent tailing. This technology also analyzes past crime statistics to measure the level of danger in a given situation. It is deemed very dangerous if such signs are detected late at night or in an area with a high crime rate. An AI surveillance camera notifies the police immediately if it concludes that there is a high risk of crime based on red flags like a person tailing while hiding his or her face with a hat and mask or in possession of a weapon.

[Soundbite] KIM GEON-WOO(ETRI) : "Advanced countries are changing their crime-fighting paradigms from counteraction to prevention. Such change is going to bring a new cutting-edge social safety system when combined with the artificial intelligence CCTV technology."

The research institute plans to boost the prediction accuracy rate by having the AI learn various crime types from court rulings before testing the system in Seocho-gu District in Seoul at the end of this year and completing the development by 2022.