A cross-university team from National Tsing Hua University and National Taiwan University won the competition with an image recognition accuracy rate of 81%.
Kuan further indicated that the competition demonstrated the innovative potential of young people to combine technology and creativity in addressing marine debris challenges.
Over several years, NAMR has developed its Marine Debris Image Dataset Platform, which contains more than 20,000 marine debris images and tens of thousands of annotated marine debris objects.
The third-place team from National Central University and National Taiwan Normal University expressed its hope of contributing to efforts to address marine debris challenges.
The 2026 International Marine Debris Image Recognition AI Challenge evaluated entries primarily on mean Average Precision (mAP) for object detection and model lightweighting performance.