Author Bio ▼

IFSEC Global is the online community for the Security and Fire industry. Our market-leading live events span the globe, connecting buyers and sellers.
April 21, 2020

Sign up to free email newsletters

Download

Whitepaper: Normal service resumed? How video technology supports our new reality

Video analytics

The use of AI in predicting future crime in South Korea

The Electronics and Telecommunications Research Institute (ETRI), based in South Korea, is in the process of developing a real-time CCTV system which will analyse past crime statistics to predict the level of crime risk. The team at IFSEC Global’s partner brand in South Korea, SECON, takes a closer look at the process.

While ‘foreseeing crime’ has been regarded as unrealistic, it is not impossible. Based on past crime statistics and CCTV footage, Korean researchers are developing an AI system which automatically analyses the probability of a crime, to prevent future incidents.

SECON-FutureCrimeAnalytics-20

ETRI is currently working on ‘Predictive Video Surveillance Core Technology’. The present situation analysis and its comparison with the past crime data, is said to enable the system to calculate the degree of danger and act against it in advance. This will help with the prediction of assaults taking place at a certain time and place, or crime details regarding criminals who may have committed violent crimes.

Video analytics and crime prediction

To aim for a more effective foreseen public order, CCTV video analysis is currently being added onto the statistical crime predication system. In other words, if a crime prediction system was merely a tool to measure the risk based on the analysis of past crimes, this new technology manages to apply real time data and predict crime risk in minutes and hours.

The disadvantage of using past crime CCTV footage is that analytics may fail to spot abnormal behaviors. To overcome this, researchers are trying to depict the current situation by referring to past crime patterns – almost like a Déjà vu of crimes committed in the past. Since finding crime prone areas with the naked eye is in many cases almost impossible, risk prediction consisting of an AI analysis procedure is thought to be highly effective.

‘Intelligent CCTV Video Analysis Technology’ enables researchers to make an accurate observation of the current situation, where certain situations can be determined by sounds or images. For example, the sound of footsteps determines whether the person is in a hurry or in a tranquil state. Visual intelligence technology offers clearer visibility on whether the person is wearing a hat, glasses, masks, carrying bags or tools on the screen.

An additional function of AI CCTV is comparing the current situation with past crimes in order to measure the level of danger. For example, if a man wearing a mask and a hat follows a woman along an obscure street at 2AM, a high alert alarm will go off. On the contrary, if the same situation were to be observed in the town centre at 2PM, the level of danger will decrease significantly.

The research being used currently consists of over 20,000 court decisions analyses and videos provided by the Florida State University, which contain crime simulations.

Furthermore, specific technology which deals with the management of sexual assault from ex-convicts will also be “thoroughly developed”. For a more distinct identification, data from ETRI’s Person Re-ID (re-identification) technology and the route of those wearing electronic anklets, can be directly transferred to the nearest CCTV system. ETRI believes this will result in more accurate management of those liable to commit crime, producing swifter actions against potential incidents.

“Our aim is to establish a futuristic cutting-edge social security system developed by a neural model, where CCTV systems will not only be able to merely perceive crimes, but also possess the ability to foresee and prevent up to 80% of crime risk,” says Kunwoo Kim, Director of Certification and Physical Security Research.

Download the Intruder Alarm Report 2020

Download this report, produced in conjunction with Texecom, to discover how increasing processing power, accelerating broadband speeds, cloud-managed solutions and the internet of things and transforming the intruder alarm market, and whether firms are adopting these innovative new technologies.

AlarmReport-Main-19

Related Topics

Subscribe
Notify of
guest
0 Comments
Inline Feedbacks
View all comments