Domestic researchers achieved first place in the international intelligent surveillance competition held at the world's largest video security conference. In particular, they swept all categories in the fields of tracking and detection, proving their world-class technology.
The Electronics and Telecommunications Research Institute (ETRI) announced on the 3rd that it achieved this result at the Intelligent Surveillance International Competition (PETS 2025 Challenge) held from the 11th for three days in Tainan, Taiwan.
This competition, organized by the world's largest video security conference (AVSS), is an evaluation competition for intelligent surveillance technology hosted by the University of Reading in the UK and sponsored by the European Border Security Cooperation Project (EUMARS). It comprehensively evaluates technologies that recognize and track people, vehicles, and ships based on various sensor images such as red, green, and blue (RGB), thermal imaging, short-wave infrared (SWIR), and ultraviolet (UV). Challenging conditions are given, similar to actual surveillance environments, including differences in sensor resolution, lighting and weather changes, and complex object congestion.
In this online competition, participating teams analyzed video data and submitted results that were automatically evaluated for performance using internationally standardized metrics such as Multiple Object Tracking Accuracy (MOTA), Identification F1 Score (IDF1), and Higher Order Tracking Accuracy (HOTA). As a result, ETRI researchers secured first place in the tracking category, with a joint research team from China and the UK capturing second place. In the detection category, National Yang Ming Chiao Tung University in Taiwan took first place, while ETRI achieved second.
ETRI researchers implemented an artificial intelligence (AI) based system that integrates the entire process from video recognition to detection and tracking. Thanks to this, they maintained high accuracy even in complex environments and responded stably to problems commonly encountered in actual surveillance systems, such as small objects, differences in sensor resolution, and frame drops.
The researchers included Kim Sang-won, Kim Dong-young, Lee Kyung-oh, and Kim Kwang-joo, affiliated with ETRI's Daegu-Gyeongbuk Research Headquarters. Kim Dong-young, a postdoctoral researcher, presented a digital twin-based pixel-level wildfire early detection technology during the official session of AVSS.
The researchers achieved this result by advancing intelligent transportation control and video analysis technologies accumulated over 30 years through national research and development projects, as well as object tracking algorithms developed by predecessor researchers. This technology is expected to be applicable in various fields, including urban traffic monitoring, port security, and maritime surveillance. It is particularly evaluated as a key technology for the next-generation smart city surveillance system, as it operates stably despite changes in weather and lighting.
ETRI is currently in discussions for technology transfer with domestic security and video solution companies, smart city platform operators, and manufacturers of port and maritime surveillance systems, with the goal of commercialization by 2027. If commercialized, it is expected to enhance public safety and urban operational efficiency in areas such as early traffic accident detection, traffic flow optimization, illegal vessel monitoring, and responses to maritime distress and pollution incidents.
ETRI Daegu-Gyeongbuk Research Headquarters Director Byun Woo-jin said, "This is an achievement that proves the excellence of our technology, which operates stably in various lighting and weather conditions, distances, object sizes, and obstacle environments, on a global scale," adding, "We will continue to strengthen practical research and development to contribute to public safety and the creation of social value."