Associate Professor
Computer Vision Laboratory
Dept. of Computer Science and Engineering
POSTECH (Pohang University of Science and Technology)
Pohang, Korea


CVPR 2016:
(03/2016) We have four papers accepted to CVPR 2016 (2 orals, 1 spotlight and 1 poster).

DPPnet for Image Question Answering:
(01/2016) The sourcecode of DPPnet paper is released in the project page.
(11/2015) We just released our arXiv paper about image question answering. This is a very unique and interesting approach based on dynamic parameter prediction for CNN. Check our paper.

Naver Faculty Award:
(01/2016) I received the Naver young faculty award.

MDNet for Visual Tracking:
(01/2016) The sourcecode of MDNet is released. Visit the project page.
(12/2015) I am very proud that MDNet won The Visual Object Tracking Challenge 2015 (VOT2015)! MDNet showed the best performance among 62 submitted or tested algorithms in the challenge. Read the result paper (Table 1 for summary) for the challenge details.
(10/2015) We obtained outstanding results of visual tracking using deep learning. Check our arXiv paper for details.

TransferNet: Transfer Learning for Semantic Segmentation:
(01/2016) The project page for TransferNet is open. Visit here.
(12/2015) Our paper about transfer learning for semantic segmentation is now available. Read our arXiv paper for details.

AAAI 2016:
(11/2015) We have a paper accepted to AAAI 2016.

NIPS 2015:
(09/2015) Our DecoupledNet paper has been accepted for a spotlight presentation at NIPS 2015. The acceptance rate of oral+spotlight presentations is only about 4.5% (82 out of 1858 submissions).
(03/2015) I will serve as an area chair for NIPS 2015.

ICCV 2015:
(09/2015) We have two papers accepted to ICCV 2015.
(02/2015) I will serve as an area chair for ICCV 2015.

(08/2015) I was appointed as an Area Editor for Computer Vision and Image Understanding Journal.

(08/2015) A paper about joint image clustering and labeling is accepted to TPAMI.

DecoupledNet for Semi-Supervised Semantic Segmentation:
(06/2015) The code for DecoupledNet is available in our project page.
(06/2015) We obtained excellent results from our new semi-supervised semantic segmentation algorithm with heterogeneous annotations. Refer to our paper and project page. [Paper] [Project page]

Semantic Segmentation on PASCAL VOC 2012 Dataset:
(06/2015) Source code and model of our deconvolution network are available in our project page! [Project page]
(05/2015) Our paper entitled "Learning deconvolution network for semantic segmentation" is now available at arXiv. [link]
(04/2015) I am excited that our deconvolution network achieved 72.5%, which is the best among the algorithms trained only on PASCAL VOC dataset (the 3rd overall). See the leaderboard for details.

ICML 2015:
(04/2015) We have a paper accept to ICML 2015.

Beyond Chain Models for Visual Tracking: a Trilogy:
(02/2015) We constructed project page for our recent works in visual tracking based on unconventional graphical models, which include orderless tracking, tracking by sampling graphical models, and online graph-based tracking. [Project page]

MSRA Fellowship Award:
(09/2014) Seunghoon won the prestigeous Microsoft Research Asia Fellowship award. See MSRA page for details.

Research Interests

Computer vision, machine learning, deep learning


  • Ph.D. in Computer Science, University of Maryland, College Park, MD, USA, 12/2005
  • M.S. in Computer Engineering, Seoul National University, Seoul, Korea, 08/2000
  • B.S. in Computer Engineering, Seoul National University, Seoul, Korea, 02/1997

Selected Recent Publication [More]

Project Pages with Codes, Datasets and Results

  • Fast nearest neighbor search: [Paper], [Project]
  • Tracking with occlusion reasoning: [Paper], [Project]
  • Event detection: [Paper], [Project]
  • Generalized background subtraction (ECCV 2014): [Paper], [Code]
  • Generalized background subtraction (ICCV 2011): [Paper], [Project]
  • Joint human segmentation and pose tracking: [Paper], [Dataset]
  • Beyond chain models for visual tracking: [Project]
  • DeconvNet (deconvolution network for semantic segmentation): [Project]
  • DecoupledNet (semi-supervised semantic segmentation): [Project]
  • CNN-SVM (ICML 2015) [Project]
  • OGBDT tracking (ICCV 2015) [Project]


Professional Service

  • Conference/workshop organizer: DTCE 2012 (organizing chair), ACCV 2014 (demo chair)
  • Area chair: NIPS 2015, ICCV 2015, ACCV (2012, 2014, 2016), WACV 2014, ACML 2016
  • Program committee: NIPS 2016, CVPR (2007~2016), ICCV (2007, 2009, 2011, 2013), ECCV (2010, 2012, 2014), ACCV (2009, 2010), AAAI (2016), IJCAI (2013), 3DV (2015), ICME (2013, 2014), PSIVT (2013)
  • Journal Associate Editor: Computer Vision and Image Understanding, Machine Vision and Applications, IPSJ Trans. on Computer Vision and Applications
  • Journal reviewer: TPAMI, IJCV, CVIU, TIP, IVC, TCSVT, TSP, TNNLS, T-IFS, JVCI, etc.

Selected Talks and Presentations [More]

  • Deep Weakly Supervised Learning in Computer Vision, KIAS (04/2016)
  • Learning Deconvolution Network for Semantic Segmentation, KAIST (04/2016), Yonsei University (03/2016), Naver (01/2016), Computer Vision Seminar at Univ. of Maryland, CSE Colloquium at Penn. State University, VASC Seminar in Robotics Institute at Carnegie Mellon University, GRASP Seminar at Univ. of Pennsylvania (10/2015), Willow Team at INRIA (07/2015), SNU (06/2015), Machine Learning Center Workshop at KCC (06/2015), UNIST (05/2015)
  • Deep Learning for Visual Recognition, SAIT (02/2016)
  • Deep Learning Architectures in Computer Vision Applications, EDA Workshop (02/2016)
  • Deep Learning for Visual Question Answering, IPIU2016 Tutorial (02/2016)
  • Structured Prediction using Convolutional Neural Networks, Deep Learning Workshop at SK Telecom (10/2015)
  • Deep Learning at POSTECH Computer Vision Lab., FCS Lab. at Samsung Electronics (10/2015), Hyundai Motors (08/2015)
  • Deconvolutions in Convolutional Neural Networks, Pattern Recognition and Machine Learning Summer School (08/2015)
  • Combinatorial Optimization in Computer Vision, IEEK Image Understanding Tutorial (08/2015)
  • Structured Prediction by SVM for Computer Vision Applications, Yonsei University (05/2013, 05/2015), IEEK Image Understanding Tutorial (08/2013, 11/2013), POSTECH (03/2013)
  • Deep Learning: Technologies and Applications, ImageNext (05/2015), SK-Planet (03/2015)
  • Pedestrian Detection: Shallow and Deep Learning, ETRI (01/2015)
  • Beyond Chain Models for Visual Tracking, SAIT (12/2014), ACCV Area Chair Workshop at NTU (09/2014), KCCV at SNU (08/2014)
  • Machine Learning for Visual Tracking, IEEK Image Understanding Tutorial (08/2014)
  • Spectral Clustering, Pattern Recognition and Machine Learning Summer School (PRMLSS) at Yonsei University (08/2014)
  • Tracking and Video Segmentation for Sports Broadcasting, The Korean Society of Broadcast Engineer Workshop (04/2014)
  • Recursive Bayesian Estimation: Applications to Visual Tracking, Pattern Recognition and Machine Learning Winter School (PRMLWS) at Yonsei University (04/2012)
  • Generalized Background Subtraction, SNU MAE (05/2012), DGIST (04/2011), SNU EE (05/2011), ETRI (07/2011), Hanyang University (01/2013, 01/2014)

Valid XHTML 1.0! Valid CSS!