I am a 1st year Ph.D. student in Artificial Intelligence (AI) program at Seoul National University, advised by Prof. Byoung-Tak Zhang. My research interests mainly span in the area of machine learning, computer vision, and natural language processing. More specifically, I have focused on machine learning algorithms for AI agent that can (1) see everyday scenes, (2) communicate with human via natural language, and (3) make decisions using vision & language information.
NEW! [August 2020] I’m starting my Ph.D. in this fall.
NEW! [June 2020] From July, I’ll join SNU AI Institute (AIIS) as a researcher.
show moreNEW! [January 2020] Our paper has been accepted to ICASSP 2020!
NEW! [December 2019] From January, I'll be a research intern at SK T-Brain!
NEW! [November 2019] I gave a spotlight talk at Video Turing Test workshop, ICCV 2019.
NEW! [October 2019] I gave an invited talk at SK Telecom AI Center.
NEW! [August 2019] Excited to announce that our paper has been accepted to EMNLP 2019.
NEW! [June 2019] Our proposed method ranks 3rd place in Visual Dialog Challenge 2019!!
NEW! [August 2018] We have a paper accepted to ECCV 2018 Workshop on VizWiz Grand Challenge.
|DialGraph: Sparse Graph Learning Networks for Visual Dialog|
Gi-Cheon Kang, Junseok Park, Hwaran Lee, Byoung-Tak Zhang, Jin-Hwa Kim
|Label Propagation Adaptive Resonance Theory for Semi-Supervised Continuous Learning|
Taehyeong Kim, Injune Hwang, Gi-Cheon Kang, Won-Seok Choi, Hyunseo Kim, Byoung-Tak Zhang
|Dual Attention Networks for Visual Reference Resolution in Visual Dialog|
Gi-Cheon Kang, Jaeseo Lim, Byoung-Tak Zhang
Paper Code Slides
|Contextualized Bilinear Attention Networks|
Gi-Cheon Kang, Seonil Son, Byoung-Tak Zhang
ECCV Workshop on VizWiz Challenge 2018
- Dual Attention Networks for Visual Reference Resolution in Visual Dialog
- Video Turing Test Workshop, ICCV 2019 (Spotlight Talk)
- T.TOC, SK Telecom AI Center
|Fast-Face Android Star|
Fast Face is an android application which detects facial landmark. It detects 68 landmarks of human face chin to eyebrow in real-time. Also, it can detect people up to 3 if you guys show your frontal faces.