SNU Professor Lee Kyung-Moo’s team won Challenge on 3D Hand Pose Estimation
- dramatically improving hand pose recognition accuracy with deep learning algorithm


▲ SNU Professor Lee Kyung-Moo’s team who won Challenge on Hand Pose Estimation 2017
Moon Kyung-Sik (Left), Lee Kyung-Moo (right)

 
On October 30^th, SNU College of Engineering (Dean Cha Kook-Heon) announced that SNU Department of Electrical & Computer Engineering Professor Lee Kyung-Moo’s team (Moon Kyung-Sik, KwangWoon Univ. Professor Jang Joo-Yong) has won 2017 Challenge on Hand Pose Estimation, which is a workshop of ICCV (International Conference on Computer Vision), held in Venice, Italy.
Prof. Lee’s team won the contest with ‘Voxel-based 3D Hand Pose Estimation technique using Deep-Learning’ algorithm. It is a new voxel-based deep learning network and learning technique, easier but more accurate than conventional way, receiving exceptionally high score.
 
Hand Pose Challenge is a contest to accurately estimate the geometry and the important features of the hands and fingers and the important features from distance images acquired with 3D distance sensors. 15 teams from Canada, China, Japan, Taiwan and Singapore participated.
 
Hand Pose Estimation technology is a core technology of HCI (Human Computer Interface), intelligent robots, VR and AR. It can be used in computer control through hand pose, user intention understanding through hand pose analysis, automatic recognition and translation f sign language for the disabled, and medical/educational/military/entertainment through interaction with virtual objects.
 
Prof. Lee explained, “The new voxel-based 3D hand pose estimation technology based on deep-learning can accurately estimate various complex hand poses that conventional technologies could not. It has raised the quality of human-machine non-verbal communication to another level, and it can be used as a core technology that increases the competitiveness of next-generation industries like display, AR and VR through better interactions between human and machine.”
 
Prof. Lee has won the best paper award in 2017 NTIRE Image Super-Resolution Challenge with ‘Super-resolution technique using improved deep-learning’ algorithm.