KETI Wins First Place at International AI Image Detection Competition

Awarded at LPCVC during CVPR 2026 ECV Workshop
Recognized for Generative AI Image Detection and Low-Power Edge Technology

A domestic research team has taken first place at an international competition with technology that distinguishes and detects AI-generated images from real ones.

KETI researchers are posing for a commemorative award photo. Photo by KETI

The Korea Electronics Technology Institute (KETI) announced on the 11th that researchers from its Mobility Platform Research Center won first place in the AI-generated image detection category of the "IEEE Low Power Computer Vision Challenge (LPCVC)" held during the ECV Workshop at "CVPR 2026," a world-renowned computer vision academic conference, which took place from June 3 to June 7 in Denver, United States.

CVPR is a leading global conference in the fields of computer vision and pattern recognition. The "IEEE Low Power Computer Vision Challenge," hosted by IEEE and sponsored by Qualcomm, is an international competition that evaluates both the performance and energy efficiency of computer vision technologies under limited power and computational resources.

AI-generated image detection is a technology field that differentiates and identifies images created by generative AI from real images. This award is significant not only because the KETI team accurately detected fake images created by generative AI, but also because their AI model’s reasoning and inference capabilities were comprehensively validated in a low-power edge device environment.

As generative AI becomes more widespread, the importance of technologies for detecting image manipulation and false content is growing. However, for these technologies to be applied in industrial settings, low-power operation, fast inference, and explainability of the results are essential.

Senior researchers Yoo Jaewoong and Park Jinman, along with researcher Ganzorig from the KETI Mobility Platform Research Center, achieved first place by developing a lightweight AI-generated image detection model, optimizing it for low-power inference, and designing a model structure capable of explaining its decision-making process.

Park Busik, Director of the KETI Mobility Platform Research Center, stated, "This award is a recognition on the global stage of the excellence of the center's VLM-based recognition and decision-making technology and our low-power inference technology. We will continue to further advance explainable edge technologies so they can be utilized in various industrial fields such as autonomous driving and robotics."

This content was produced with the assistance of AI translation services.