SNU Computer Vision Lab,
won the image super-resolution challenge

-Use ‘enhanced deep-learning high resolution algorithm’
-Take first place in all 6 categories … won the competition and won the best paper prize


▲ SNU computer vision lab led by Prof. Kyoungmu Lee:
(from the left) Bee Lim, Prof. Lee, Sanghyun Son, Heewon Kim, Seoungjun Na
 

Seoul National University (Dean Kunwoo Lee) announced that the computer vision lab led by Prof. Kyoungmu Lee won the NITRE 2017 Image Super-Resolution Challenge, which was held in Hawai on July 21, through enhanced super resolution technique using deep learning algorithm. Furthermore, they won the Best Paper Prize at the same time.

The competition was held as part of a workshop on "Computer Vision and Pattern Recognition (CVPR) 2017" that is the world's largest conference on computer vision and machine learning. The goal of the competition is to restore the low-resolution images to ultra-high-resolution images using software. The competition was supported by Google, NVIDIA, and Twitter.

More than 100 teams from all over the world participated in the contest, of which 20 teams from 8 countries (Hong Kong Central University, University of Illinois, USA, Peking University, Tsinghua University, Harbin Institute of Technology, Sense Time, Finland Tempe University, India IIT, Swiss EPFL ) participated in the finals.

Professor Lee's team implemented a dramatically new deep-learning neural network and optimized it for ultra-high-resolution image restoration. This was the only winner in all six categories, won the competition by a large margin.

The ultrahigh-resolution image restoration technology is the core technology of next-generation image processing. It can improve the performance of objects and people recognition, and high-dimensional image analysis in various fields such as autonomous vehicles, intelligent robots, surveillance cameras and so on.

Prof. Lee said, "With the new deep-learning-based ultra-high resolution algorithm, we can restore 8x and 16x magnification image to the almost original image. It is meaningful since high-price equipment performance can be achieved by low-price equipment, and also it can be used in a variety of fields."

Meanwhile, Prof. Lee is a leading researcher in the field of computer vision and machine learning. He is currently the vice-editor of IEEE TPAMI (Transactions on Pattern Analysis and Machine Intelligence), which is a world-famous journal on pattern recognition and artificial intelligence. He is also the chairman of organizing committee of ICCV 2019 (International Conference on Computer Vision 2019).