Byungsoo Ko
Research engineer
@ NAVER & LINE corp.
kobiso62 [at] gmail.com
π linkedin | π github | π¨βπ google scholar | π kaggle | π‘ blog
Research Interests
Computer vision
Multi-modal contrastive learning
Metric learning (generalization, image retrieval)
Line segment detection
Text-to-image generation (especially, transformer-based approach like DALL-E)
Publications
(* authors contributed equally)
Granularity-aware Adaptation for Image Retrieval over Multiple Tasks
Learning with Memory-based Virtual Classes for Deep Metric Learning
Byungsoo Ko*, Geonmo Gu*, Han-Gyu Kim
International Conference on Computer Vision (ICCV), 2021
[paper] [code] [presentation video] [ppt] [poster]
RTIC: Residual Learning for Text and Image Composition using Graph Convolutional Network
Proxy Synthesis: Learning with Synthetic Class for Deep Metric Learning
Geonmo Gu*, Byungsoo Ko*, Han-Gyu Kim
The Association for the Advancement of Artificial Intelligence (AAAI), 2021
[paper] [code] [presentation video] [ppt] [poster]
An Effective Pipeline for a Real-world Clothes Retrieval System
Yang-Ho Ji*, HeeJae Jun*, Jongtack Kim*, Youngjoon Kim*, Byungsoo Ko*, Insik Kim* Hyong-Keun Kook*, Jingeun Lee*, Sanghyuk Park* Sangwon Lee*
CVFAD Workshop on Computer Vision and Pattern Recognition (CVPR), 2020
[arxiv]
Embedding Expansion: Augmentation in Embedding Space for Deep Metric Learning
Byungsoo Ko*, Geonmo Gu*
Conference on Computer Vision and Pattern Recognition (CVPR), 2020
[arxiv] [code] [supplementary video] [ppt] [poster]
Symmetrical Synthesis for Deep Metric Learning
Geonmo Gu*, Byungsoo Ko*
The Association for the Advancement of Artificial Intelligence (AAAI), 2020
[arxiv] [code] [supplementary video] [poster] [spotlight ppt]
A Benchmark on Tricks for Large-scale Image Retrieval
Naive semi-supervised deep learning using pseudo-label
Finding a News Article Related to Posts in Social Media: The Need to Consider Emotion as a Feature
Byungsoo Ko, Chanyong Park, Dongkeon Lee, Jaewon Kim, Ho-Jin Choi, Dongsoo Han
The 1st International Workshop on Dialog Systems (IWDS), 2018
[paper] [code]
Pseudo Labeling Using Gaussian Process for Semi-supervised Deep Learning
Zhun Li, Byungsoo Ko, Ho-Jin Choi
IEEE International Conference on Big Data and Smart Computing (BigComp) , 2017
[paper]
Empathy Bot - Conversational Service for Psychiatric Counseling with Chat Assistant
Kyo-Joong Oh, DongKun Lee, Byungsoo Ko, Jonghwan Hyeon, Ho-Jin Choi
Precision Healthcare through Informatics: MEDINFO, 2017
[paper]
A Chatbot for Psychiatric Counseling in Mental Health Care Service Based on Emotional Dialogue Analysis and Sentence Generation
Neural Network-based Autonomous Navigation for a Homecare Mobile Robot
Byungsoo Ko, Ho-Jin Choi, Chansol Hong, Jong-Hwan Kim, Oh Chul Kwon, Chang D. Yoo
BigData4Healthcare (workshop in IEEE BigComp), 2017
[paper]
Honors & Awards
2nd Place in DeepFashion2 Clothes Retrieval Challenge at CVPR workshop (CVFAD), 2020
Publication: [arxiv]
8th Place in Google Landmark Retrieval in Kaggle, 2019
Publication: [arxiv]
29th Place in Google Landmark Recognition in Kaggle, 2019
Government-funded scholarship in KAIST, 2016-2017
Dean's award for graduation with honors in CNU, 2016
Exchange student scholarship in CNU, 2013
Dean's list scholarship in CNU,2010-2015
Career
Research engineer at NAVER & LINE corp., Apr. 2018 - Now
Master's student researcher at Knowledge Engineering and Collective Intelligence Lab in KAIST, Mar. 2016 - Apr. 2018
Undergraduate student researcher at Applied Scientific Computing Lab in CNU, Dec. 2014 - Dec. 2015
Mobile application developer intern in Mobidaim, Jan. 2015 - Feb. 2015
Education
Master's degree: Korea Advanced Institute of Science and Technology (KAIST), 2016 - 2018
Bachelor's degree: Chungnam National University (CNU), 2010-2016
Exchange student at Memorial University of Newfoundland (MUN) in Canada, 2014