Superb AI announced on the 21st that it has won first place in the Hanwha Systems AI Challenge for two consecutive years.
The Hanwha Systems AI Challenge is a technology competition that comprehensively evaluates the accuracy and lightweight level of artificial intelligence models that detect objects using infrared camera data.
In the 2024 competition, researcher Im Eun-soo won first place, and in the 2025 challenge, researcher Kim Seung-hyun and researcher Ko Gyeong-ryeol teamed up to take the top spot again.
Infrared data is harder to obtain than general RGB images, and is a key variable in developing a high-performance detection model. The Superb AI research team secured training data by combining a synthesis technology that automatically converts RGB images into infrared images and a technology that generates infrared images based on image description text. In addition, they expanded the dataset by applying a pseudo-labeling technique in which AI learns by estimating the correct answer to unlabeled data, and maintained detection accuracy in various environments through data augmentation technology.
Superb AI CEO Hyunsoo Kim said, “This award proves the competitiveness of our generative artificial intelligence technology and AI data processing capabilities, and we will continue to accelerate our advance into the global market through technological innovation.”
- See more related articles
You must be logged in to post a comment.