The Korea Electronics Technology Institute (KETI, President Shin Hee-dong) has proven the competitiveness of its AI image high-resolution technology by winning first place in the world competition for real-time image super-resolution based on artificial intelligence.

KETI announced on the 24th that it won first place in the international competition for real-time image super-resolution based on artificial intelligence (AI) held online in Tel Aviv, Israel on the 23rd.

KETI Intelligent Image Processing Research Center (Dr. Seong-Je Kim, Dr. Jin-Woo Jeong, and 6 others) participated in the 'Real-Time Image Super-Resolution Challenge' of the 'Mobile AI & AIM 2022 Workshop' from May to July, and recorded the highest evaluation score among 28 teams (250 people) around the world.

This workshop is a major program of the European Conference on Computer Vision (ECCV), the world's most prestigious conference in the field of AI and computer vision, and is known for covering the excellence of various image processing technologies (such as super-resolution, depth estimation, and rendering) on mobile devices.

The European Conference on Computer Vision (ECCV) is held together with the Conference on Computer Vision and Pattern Recognition (CVPR) and the International Conference on Computer Vision (ICCV).It is one of the world's top three international conferences in the field of computer vision and is being held in Tel Aviv, Israel from the 23rd to the 27th.

This challenge was conducted by developing an AI algorithm for image super-resolution using training data provided by the organizer, and then evaluating the score by measuring the accuracy of the results and the processing speed on mobile devices.

The KETI research team built an AI-based super-resolution network by stacking multiple convolutional neural network (CNN) blocks and trained the developed network to achieve excellent performance, enabling them to present "real-time image super-resolution technology that operates on mobile devices."

According to the Intelligent Image Processing Research Center, existing ultra-high-resolution technologies have made it difficult to perform real-time image processing on mobile devices, but the newly developed technology operates in real time on mobile devices and can improve image quality.

The research team plans to expand related research to improve image quality even in low-light and high-turbidity environments in the future, thereby developing various AI-based image processing technologies within mobile devices.

Kim Je-woo, director of the Intelligent Image Processing Research Center, said, “Image super-resolution technology is expected to improve the quality of existing mobile content, maximizing viewers’ immersion,” and added, “KETI will do its best to develop AI-based image recognition and analysis technology with the goal of developing original and core technologies that lead the field of image processing.”

Meanwhile, the technology was developed through the 'Intelligent Media Aspect Ratio Conversion Technology Maintaining Properties' project, which was conducted as part of the Media Intelligence Technology Development Project with the support of the Ministry of Science and ICT and the National IT Industry Promotion Agency.