SNU Ph.D. Research Student Gyeongsik Moon’s Computer Vision Laboratory Gains Attention by Publishing Four Papers on ECCV2020

- Korea's first to be recognized by an academic journal known to be within the TOP3 of all areas of artificial intelligence

- Portion regarding details related to the human pose estimation is selected as an oral presentation paper


▲ From left Professfor Kyoung Mu Lee of SNU College of Engineering Department of Electrical and Computer Engineering, Ph.D Research Student Gyeongsik Moon

 
Four papers on Human Pose Estimation with Ph.D Researcher Gyeongsik Moon of the SNU Computer Vision Laboratory (Advisor Kyung Moo Lee, Department of Electrical and Computer Engineering Professor) as the first author, have been published in ECCV (European Conference on Computer Vision) 2020, one of the best societies in the field of artificial intelligence and computer vision. In particular, one of them was selected as an oral presentation paper, and it is important to note that only a small number of influential papers are selected to be orally presented.
 
In accordance to Google Scholar's criteria, EECV is within the TOP3 of the best academic conference in all areas of artificial intelligence and is known to have high standards for publishing papers while having a great influence on them. This year, a total of 5,025 effective papers were submitted and with 1,361 being selected, they showed to have a 27 percent publication rate, and out of these, 104 of them or 2 percent were selected to be orally presented.
 
"This is the first time in Korea that a graduate student, as the main author, has published four papers simultaneously at a TOP conference on artificial intelligence, and such a case is extremely rare globally, which is proof of world-class level research work in the AI field achieved by graduate students within the country," said Professor Kyoung Mu Lee, the advisor.
 
The presentation paper of Professor Kyoung Mu Lee's research  team is as follows:
- Development of a new I2L-MeshNet network that is capable of accurate three-dimensional human pose and mesh estimation from a single RGB color image
(Original title: Gyeongsik Moon and Kyoung Mu Lee, "I2L-MeshNet: Image-to-Lixel Prediction Network for Accurate 3D Human Pose and Mesh Estimation from a Single RGB Image," ECCV202020)
 
- Development of a deep learning network based on pharmacists that restores the three-dimensional shape of a person's hand from image input.
(Original title: Gyeongsik Moon, Takaaki Shiratori, and Kyoung Mu Lee, “DeepHandMesh: A Weakly-supervised Deep Encoder-Decoder Framework for High-fidelity Hand Mesh Modeling”, ECCV2020 (Oral))
 
- Establishing a methodology and data set to estimate the three-dimensional poses of both hands interacting from a single RGB color image
(Original title: Gyeongsik Moon, Shoou-I Yu, He Wen, Takaaki Shiratori, and Kyoung Mu Lee, “InterHand2.6M: A Dataset and Baseline for 3D Interacting Hand Pose Estimation from a Single RGB Image”, ECCV2020)
 
 
- Proposal of a new graph-based convolutional neural network to restore three-dimensional human posture from two-dimensional human posture information
(Original title: Hongsuk Choi*, Gyeongsik Moon*, and Kyoung Mu Lee, “Pose2Mesh: Graph Convolutional Network for 3D Human Pose and Mesh Recovery from a 2D Human Pose”, ECCV2020 (*Joint authors)
 
ECCV2020 will run online for five days from August 23.