A domestic research team has taken the first step toward realizing 'Minority Report' by successfully developing a technology that combines CCTV and artificial intelligence technology to detect and predict signs of crime in advance.

The Electronics and Telecommunications Research Institute (ETRI) announced on the 18th that it has completed the development of 'Dejaview' technology that analyzes CCTV footage, crime statistics, and location information to detect signs of crime in advance and predict the possibility of crime occurring.

The deja vu technology developed by the research team was based on the tendency for crimes to occur repeatedly in a similar pattern to past crimes, including type, method, location, and time.

This technology is an artificial intelligence (AI) technology that measures and predicts crime risk by comparing and analyzing situations in which past crimes occurred and the current progress of the crime.

ETRI's Deja Vu technology, which probabilistically predicts crime occurrence, can be divided into two technologies depending on the target and method of application.

First, spatiotemporal crime prediction technology analyzes what type of crime is at high risk at a specific location and time.

For example, if a situation similar to a past crime pattern occurs in a quiet place where a crime has occurred late at night, the risk is considered very high. <br />
This is a method to block and prevent crimes in advance.

It is expected to be mainly used in the future by police and CCTV integrated control centers of local governments across the country.

Based on statistical information such as past crimes, AI automatically analyzes real-time CCTV footage to compare and measure similarity with the crime situation.

It can also immediately identify and track suspicious situations of crime or disaster, such as stalking, collapse, or early-stage fire, and understand the overall situation, such as recognizing the properties of pedestrians and vehicles (drugs, smuggling, etc.).

ETRI, in collaboration with Seocho-gu, Seoul, developed a crime prediction map (PCM) by analyzing big data on 32,656 CCTV incidents and accidents in the region over a three-year period from 2018.

Based on past crime statistics, it shows areas with a high probability of crime by categorizing the crime by date and time, location, and type of incident (violent crime, traffic accident, fire, etc.).

Additionally, PCM presents real-time crime risk on the screen.

The controller can see the current crime risk level in each area at a glance through the crime prediction map.

This makes it possible to conduct predictive selective control that prioritizes areas with a high risk of crime occurring.

The crime prediction performance of the crime prediction map was measured at 82.8% according to the performance test standards of the Telecommunications Technology Association (TTA) of Korea.

ETRI also developed a second Deja Vu technology, a person-centered recidivism prediction technology.

This technology is strictly limited to those subject to electronic surveillance at high risk of recidivism.

Analyze the risk based on the movement patterns of electronically monitored subjects.

The current electronic surveillance system determines whether movement restrictions are in place based on positioning information.

ETRI in the existing systemIf this technology is applied, if a person subject to electronic surveillance periodically violates movement restrictions due to work or other reasons, this can be analyzed through artificial intelligence technology, the risk of repeat offenses can be identified, and a response can be taken.

The performance of electronic surveillance subject location-based risk analysis is around 95%.

A total of 7,397 daily alerts were tested for TTA certification.

This technology can be utilized in the Ministry of Justice's Location Tracking Central Control Center.

It is expected that in the future, it will develop into an AI electronic surveillance system that comprehensively analyzes not only location information but also various daily life, interview, and medical information.

The ETRI research team plans to develop the newly developed crime sign detection and prediction source technology into a system specialized in safety services.

We plan to expand and apply it to various safety services, such as a crime prediction system tailored to each region, a preemptive response system for national infrastructure risks such as airports, energy facilities, and factories, a safety system for major national events, and a preemptive response to risky behavior (intention) for security purposes.

To achieve this, the development of world-class video security AI core technologies that can accurately analyze, understand, and infer situations occurring in real time from CCTV footage must be supported.

Based on the developed Deja Vu technology, ETRI researchers plan to provide crime sign detection and prediction services specialized for security sites in cooperation with each local government and control agency.

In addition, we plan to expand the development of AI electronic surveillance technology to recognize and respond to deviant behavior of high-risk electronic surveillance subjects in advance.

The commercialization of Deja Vu technology is expected to take place at the end of next year.

ETRI Artificial Intelligence Convergence Security Lab’s Chief Researcher Kim Geon-woo said, “CCTV goes beyond simply detecting crimes and can detect risks on its own.“We have developed a video security technology that predicts and prevents the possibility of a disaster in advance. Through this technology, we have laid the technological foundation for a leap forward to a future-oriented, cutting-edge social safety system,” he said.

This project was carried out until last year as the 'Development of Predictive Video Security Core Technology for 5G-based Preemptive Risk Response' of the Ministry of Science and ICT. It was selected as one of the 100 Best National Research and Development Projects in 2023, one of the 25 Best IITP ICT R&D Projects, and one of the 9 Representative Achievements of ETRI.