Donggeun Yoo

Co-founder & research scientist at Lunit,
PhD student in School of Electrical Engineering, KAIST.,
Google Scholar | Linkedin


I received my BS in 2011 and MS in 2013 at Electrical Engineering, KAIST. From 2013, I am a PhD student at KAIST, working on visual representation learning with Prof. In So Kweon, but I decided to take some time off to join Lunit in 2017. As a full-time research scientist at Lunit, I am devoted to developing advanced AI for medical image analysis and interpretation via cutting-edge deep learning technology. During my internship experience at Adobe Research in the US, I worked on large-scale video representation learning. My research interests include visual recognition problems with deep learning approaches.

Research Interests

Representation learning, unsupervised learning, large-scale learning, for visual recognition.

Awards | Competitions

Mar. 2017. 

Our transfer learning method, Multi-Scale Pyramid Pooling (MPP), was employed to Samsung Galaxy S8 Bixby Vision for fine-grained object classification and product retrieval.
Related paper

Dec. 2015. 

ImageNet Large Scale Visual Recognition Challenge (ILSVRC)
5th place at the classification and localization task among 23 participants including world leading companies such as Google, Microsoft, Samsung and Qualcomm.
Invited to ILSVRC Workshop in ICCV 2015 to provide a talk about "Multi-Class AttentionNet", which was selected as one of top 3 novel localization approaches.
Team name: Lunit-KAIST.
Result | Workshop | Slide

Feb. 2009. 

Grand Prize in KAIST Undergraduate Research Program.
Topic: Portable Noncontact Heartbeat Sensor Using LC Oscillation.
Advisor: Prof. Songcheol Hong.


D. Kim, D. Cho, D. Yoo, I. S. Kweon,
Two-Phase Learning for Weakly Supervised Object Localization,
IEEE International Conference on Computer Vision (ICCV), 2017.
Paper (preprint)

D. Yoo, S. Park, K. Paeng, J.-Y. Lee, I. S. Kweon,
Action-Driven Object Detection with Top-Down Visual Attentions,
arXiv preprint, 2016.
Paper | Video - person, multi

D. Yoo, N. Kim, S. Park, A. S. Paek, I. S. Kweon,
Pixel-Level Domain Transfer,
European Conference on Computer Vision (ECCV), 2016.
Paper | Supp | LookBook dataset | Code written by Fei Xia
(This paper was also invited to TASK-CV)

D. Yoo, S. Park, J.-Y. Lee, A. S. Paek, I. S. Kweon,
AttentionNet: Aggregating Weak Directions for Accurate Object Detection,
IEEE International Conference on Computer Vision (ICCV), 2015.
Paper | Supp

Y. Yoon, H.-G. Jeon, D. Yoo, J.-Y. Lee, I. S. Kweon,
Learning a Deep Convolutional Network for Light-Field Image Super-Resolution,
IEEE International Conference on Computer Vision (ICCV) - CPCV Workshop, 2015.

D. Yoo, S. Park, J.-Y. Lee, I. S. Kweon,
Multi-scale Pyramid Pooling for Deep Convolutional Representation,
International Conference on Computer Vision and Pattern Recognition (CVPR) - DeepVision Workshop, 2014.
Paper | Supp
(This method was employed to Samsung Galaxy S8 Bixby Vision)

D. Yoo, K. Paeng, S. Park,J. Lee, S. Paek, S.-E. Yoon, I. S. Kweon,
PRISM: A System for Weighted Multi-Color Browsing of Fashion Products,
International World Wide Web Conference Companion (WWW), 2014.