This composite file image shows CCTV equipment set up in Seoul’s Han River parks. (Yonhap)
GWANGJU, April 14 (Korea Bizwire) — The Korea Photonics Technology Institute announced that it has developed an artificial intelligence (AI)-based situation recognition solution that can monitor dangerous criminal and emergency situations using the sounds generated in CCTV blind spots.
The technology was developed to sense all kinds of sounds, including the ones mixed in a composite manner, going beyond the limit of existing technologies that can recognize only two to 10 different kinds of sounds associated with specific expressions like screams.
The new technology can recognize 76 different kinds of sounds, including fire alarms, footsteps as well as Korean, English and Chinese.
Within three seconds after recognizing sounds, it can sort out them with an accuracy rate of up to 91 percent.
The new solution can individually recognize sound objects that exist compositely in the real world while assessing on-site situations based on the distribution of sounds over time.
Accordingly, it can be used solely or together with video information for a physical security control system.
M. H. Lee (mhlee@koreabizwire.com)
