Gabyong Park

Ph.D ,  

NAVER Cloud

E-mail: baeckgoo@gmail.com

[CV] [google scholar]

Research Interests

My research interest is pose estimation of a fast-moving hand for interaction with virtual objects in AR/VR.

Previously, I tried to solve the problem with the following approaches:

1) Hybrid approach (classification and model-based tracking).

2) Sensor-fusion tracking (a depth camera and a gyro sensor).

3) Domain transfer learning (depth to IR).

The works were presented in ISMAR/VR conferences.


Currently, I'm interested in the following works:

- 3D hand mesh recovery from a single RGB image in the presence of motion blur. 

- Human mesh recovery from RGB-D images for remote collaboration.

- 3D face mesh recovery from a single RGB image


Main works:

BlurHand: 3D Hand Mesh Recovery in Motion-blurred RGB Images (In progress)

Gabyong Park, Tae-Kyun Kim, Woontack Woo  

(Patent: 10-2021-0097571, Republic of Korea)

3D Hand Pose Estimation with a Single Infrared Camera via Domain Transfer Learning

Gabyong Park, Tae-Kyun Kim, Woontack Woo

ISMAR 2020 (best paper nominee, 8 papers out of 302 submissions) [paper][code]

Bare-hand Depth Inpainting for 3D Tracking of Hand Interacting with Object

Woojin Cho, Gabyong Park, Woontack Woo

ISMAR 2020 [paper]

3D Hand Tracking in the Presence of Excessive Motion Blur

Gabyong Park, Antonis Argyros, Juyoung Lee, Woontack Woo

TVCG 2020 (also presented in VR 2020) [paper] [dataset]

Hybrid 3D Hand Articulations Tracking Guided by Classification and Search Space Adaptation

Gabyong Park, Woontack Woo

ISMAR 2018  [paper] 

Efficient 3D Hand Tracking in Articulation Subspaces for the Manipulation of Virtual Objects

Gabyong Park, Antonis Argyros, Woontack Woo

CGI 2016 [paper]

Hand Tracking with a Near-Range Depth Camera for Virtual Object Manipulation in an Wearable Augmented Reality

Gabyong Park, Taejin Ha, Woontack Woo

HCII 2014 [paper]