It will be helpful to small video-related businesses

ETRI (Korea Electronics and Telecommunications Research Institute) announced on the 1st that it is embarking on the development of a cloud-based intelligent video security incubation platform. This technology is expected to significantly contribute to the establishment of a future-oriented advanced security platform, as well as strengthen the market competitiveness of small video-related companies by enabling them to secure intelligent core technologies, by learning and providing intelligent security models optimized for various security environments based on a cloud platform.

While existing CCTVs are widely utilized to provide crucial evidence in solving crimes, there are many situations where accurate visual identification is difficult. As a single person monitors footage from dozens of CCTV channels for extended periods, there are instances where dangerous situations go unnoticed even when they occur.

ETRI is developing futuristic advanced policing technology that integrates artificial intelligence into security CCTVs to automatically detect dangerous situations, such as traffic accidents and crimes, in real time and notify the police, and to automatically identify and track suspects and vehicles involved in accidents.



The research team introduced the source technology possessed by ETRI for the development of this technology. That is, video deep learning technology for automatic detection of traffic accidents, re-identification technology for identifying and tracking suspects and suspect vehicles, big data learning technology for outdoor CCTV demonstration videos, and video security infringement prevention technology are utilized as foundational technologies.

Based on this core technology, we plan to infuse intelligence into public security CCTVs by developing intelligent license plate recognition technology (Deep Resolution) that not only immediately detects crimes or traffic accidents occurring during vulnerable hours, such as late at night, and accurately recognizes and tracks subsequent dangerous situations, but also overcomes the limitations of human visual perception.

ETRI plans to enable blurry, low-resolution license plates to be zoomed in to three levels of high-resolution images by the middle of next year through research and development starting this year. Additionally, by the end of next year, it will research technology that detects traffic accidents within three seconds of their occurrence and sounds an alarm. In the third and final year, we plan to lay the foundation for the commercialization of advanced public security by conducting a pilot project jointly with the National Police Agency in the Jeju Island region, where traffic accidents and crime rates are high due to the large number of domestic and international tourists and rental car usage.

ETRI's foundational technologies, such as video deep learning technology and video security intrusion prevention technology

The research team anticipates that through this technology development, it will be possible to automatically extract vehicle type, color, and model from existing CCTVs using artificial intelligence, and to determine whether a criminal is the same person as they move.

In addition, it was explained that real-time automatic recognition and tracking of security risk situations within HD-quality CCTVs collected at 30 frames per second would also be possible.

Therefore, if the intelligent technology developed by ETRI researchers is subsequently integrated into CCTV control centers operated by local governments nationwide, the National Police Agency, and unmanned security companies, it will be possible to transition to a future-oriented advanced public security platform focused on real-time response and prevention rather than post-incident recovery.

Kim Gun-woo, Project Leader at ETRI’s Information Security Research Division, stated, “We expect to lay the groundwork for future advanced policing and lead global social safety technology trends by generating research results on security intelligence based on empirical big data through R&D involving security stakeholders such as the National Police Agency and local governments, and applying them to pilot services.”