D
ahun Kim [cv]

        Ph.D. Candidate

     Robotics and Computer Vision Laboratory

     Korea Advanced Institute of Science and Technology (KAIST)

        mcahny [at] kaist.ac.kr
        mcahny01 [at] gmail.com







Research Interests
  • Deep Learning
    •  Weakly-supervised learning  
    •  Self-supervised representation learning 
    •  Video understanding

  • Computer Vision
    •  Semantic segmentation 
    •  Object recognition
    •  Video understanding
    •  Video inpainting
    •  Scene graph generation


International Publications


2019

  • Deep Video Inpainting
    Dahun Kim*
    , Sanghyun Woo*, Joon-Young Lee, In So Kweon
    IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2019, To appear

  • Deep Blind Video Decaptioning by Temporal Aggregation and Recurrence
    Dahun Kim*, Sanghyun Woo*, Joon-Young Lee, In So Kweon
    IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2019, To appear


  • Self-Supervised Video Representation Learning with Space-Time Cubic Puzzles
  • Dahun Kim, Donghyeon Cho, In So Kweon
    Association for the Advancement of Artificial Intelligence (AAAI) Jan 2019, Oral






  • Discriminative Feature Learning for Unsupervised Video Summarization 

    Yunjae Jung, Donghyeon Cho, Dahun Kim, Sanghyun Woo, In So Kweon
    Association for the Advancement of Artificial Intelligence (AAAI) Jan 2019, Oral




2018
  • LinkNet: Relational Embedding for Scene Graph
    Sanghyun Woo*, Dahun Kim*, Donghyeon Cho, In So Kweon
    Neural Information Processing Systems (NIPS), Dec 2018          




  • Learning Image Representations by Completing Damaged Jigsaw Puzzles
    Dahun Kim, Donghyeon Cho, Donggeun Yoo, In So Kweon
    IEEE Winter Conference on Computer Vision (WACV), Mar 2018, Oral




2017
  • Two-Phase Learning for Weakly Supervised Object Localization
    Dahun Kim, Donghyeon Cho, Donggeun Yoo, In So Kweon
    IEEE International Conference on Computer Vision (ICCV), Oct 2017




International Patent

  • Method and Apparatus for Hierarchical Learning of a Neural Network Based on Weakly Supervised Learning
    PCT/KR2017/013003, Pending. Nov 2017.


Domestic Publications (Korean)

2015
  • 물체후보 분할과 제안을 위한 딥러닝
    Dahun Kim, Tae-Hyun Oh, In So Kweon
    멀티미디어학회 추계학술발표대회, November 2015




Educations
  • Ph.D., KAIST - Electrical Engineering                                                                                                                                         2018.03 - present
    Advisor: prof. In So Kweon

  • M.S., KAIST - Electrical Engineering                                                                                                                                          2016.03 - 2018.02
    Advisor: prof. In So Kweon

  • B.S., KAIST - Electrical Engineering                                                                                                                                           2012.02 - 2016.02 


  • High School Diploma, Korea Science Academy                                                                                                                         2009.03 - 2012.02