Jun-Hyuk Kim

Research Scientist

junhyuk.kim@yonsei.ac.kr

I am a research scientist at Samsung Advanced Institute of Technology (SAIT). I received my Ph.D. degree from Yonsei University, South Korea, in 2022, under the supervision of Prof. Jong-Seok Lee. I am interested in making visual data more valuable in various aspects based on deep learning. In particular, I have focused on developing learned image restoration and compression models for better visual quality and less capacity, respectively.

Experiences

  • Research scientist at SAIT (Oct. 2022 - Now)
  • Research intern at NAVER AI Lab (Jun. 2021 - Nov. 2021)

  • News

    Nov. 2022 1 paper accepted in AAAI'23 (oral presentation)
    Jul. 2022 1 paper accepted in Neurocomputing
    Mar. 2022 1 paper accepted in CVPR'22
    Jul. 2021 1 paper accepted in ICCV'21
    Sep. 2020 1 paper accepted in ACCV'20
    Jul. 2020 1 paper accepted in MM'20 (oral presentation)
    Mar. 2020 1 paper accepted in Neurocomputing
    Jan. 2020 2 papers accepted in ICASSP'20 (1 oral presentation)

    Publications

  • Journals
  • Successive learned image compression: Comprehensive analysis of instability
    J.-H. Kim, S. Jang, J.-H. Choi, J.-S. Lee
    Neurocomputing, vol. 506, pp. 12-24, Sep. 2022
    Detail

    Volatile-nonvolatile memory network for progressive image super-resolution
    J.-H. Choi, J.-H. Kim, M. Cheon, J.-S. Lee
    IEEE Access, vol. 9, pp. 37487-37496, Mar. 2021
    Detail / GitHub

    Prediction of car design perception using EEG and gaze patterns
    S.-E. Moon, J.-H. Kim, S.-W. Kim, J.-S. Lee
    IEEE Transactions on Affective Computing, vol. 12, no. 4, pp. 843-856, 2021
    Detail

    MAMNet: Multi-path adaptive modulation network for image super-resolution
    J.-H. Kim, J.-H. Choi, M. Cheon, J.-S. Lee
    Neurocomputing, vol. 402, pp. 38-49, Aug. 2020
    Detail / arXiv / GitHub

    Deep learning-based image super-resolution considering quantitative and perceptual quality
    J.-H. Choi, J.-H. Kim, M. Cheon, J.-S. Lee
    Neurocomputing, vol. 398, pp. 347-359, Jul. 2020
    Detail / arXiv / GitHub

  • Conferences
  • Demystifying randomly initialized networks for evaluating generative models
    J. Lee, J.-H. Kim, J.-S. Lee
    AAAI Conference on Artificial Intelligience (AAAI), Feb. 2023
    arXiv

    Deep image destruction: Vulnerability of deep image-to-image models against adversarial attacks
    J.-H. Choi, H. Zhang, J.-H. Kim, C.-J. Hsieh, J.-S. Lee
    International Conference on Pattern Recognition (ICPR), Aug. 2022
    arXiv

    Joint global and local hierarchical priors for learned image compression
    J.-H. Kim, B. Heo, J.-S. Lee
    IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 5992-6001, Jun. 2022
    Detail / arXiv

    Just one moment: Structural vulnerability of deep action recognition against one frame attack
    J. Hwang, J.-H. Kim, J.-H. Choi, J.-S. Lee
    IEEE International Conference on Computer Vision (ICCV), pp. 7668-7676, Oct. 2021
    Detail / arXiv

    Edge attention network for image deblurring and super-resolution
    J.-W. Han, J.-H. Choi, J.-H. Kim, J.-S. Lee
    IEEE International Conference on Systems, Man, and Cybernetics (SMC), pp. 2393-2398, Oct. 2021
    Detail

    NTIRE 2021 challenge on image deblurring
    S. Nah et al. (including J.-H. Kim)
    IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, Jun. 2021
    Detail / arXiv

    Adversarially robust deep image super-resolution using entropy regularization
    J.-H. Choi, H. Zhang, J.-H. Kim, C.-J. Hsieh, J.-S. Lee
    Asian Conference on Computer Vision (ACCV), Nov. 2020
    Detail

    Multi-scale adaptive residual network using total variation for real image super-resolution
    K.-H. Ahn, J.-H. Kim, J.-H. Choi, J.-S. Lee
    International Conference on Consumer Electronics Asia (ICCE-Asia), pp. 349-352, Nov. 2020
    Detail

    Efficient bokeh effect rendering using generative adversarial network
    M.-S. Choi, J.-H. Kim, J.-H. Choi, J.-S. Lee
    International Conference on Consumer Electronics Asia (ICCE-Asia), pp. 404-408, Nov. 2020
    Detail

    Instability of successive deep image compression
    J.-H. Kim, S. Jang, J.-H. Choi, J.-S. Lee
    ACM Multimedia (MM), pp. 247-255, Oct. 2020
    Detail

    LarvaNet: Hierarchical super-resolution via multi-exit architecture
    G.-W. Jeon, J.-H. Choi, J.-H. Kim, J.-S. Lee
    European Conference on Computer Vision (ECCV) Workshops, Aug. 2020
    Detail

    AIM 2020 challenge on efficient super-resolution: methods and results
    K. Zhang et al. (including J.-H. Kim)
    European Conference on Computer Vision (ECCV) Workshops, Aug. 2020
    arXiv

    AIM 2020 challenge on real image super-resolution: methods and results
    P. Wei et al. (including J.-H. Kim)
    European Conference on Computer Vision (ECCV) Workshops, Aug. 2020
    arXiv

    Efficient deep learning-based lossy image compression via asymmetric autoencoder and pruning
    J.-H. Kim, J.-H. Choi, J. Chang, J.-S. Lee
    IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 2063-2067, May 2020
    Detail

    SRZoo: An integrated repository for super-resolution using deep learning
    J.-H. Choi, J.-H. Kim, J.-S. Lee
    IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 2508-2512, May 2020
    Detail / arXiv / GitHub

    Evaluating robustness of deep image super-resolution against adversarial attacks
    J.-H. Choi, H. Zhang, J.-H. Kim, C.-J. Hsieh, J.-S. Lee
    IEEE International Conference on Computer Vision (ICCV), pp. 303-311, Oct. 2019
    Detail / arXiv

    Deep learning-based super-resolution for digital comics
    J.-H. Kim, J.-H. Choi, C.-H. Seo, J. Chang, J.-S. Lee
    SIGGRAPH Asia 2018 Posters, Article 19, Dec. 2018
    Detail

    Generative adversarial network-based image super-resolution using perceptual content losses
    M. Cheon, J.-H. Kim, J.-H. Choi, J.-S. Lee
    European Conference on Computer Vision (ECCV) Workshops, pp. 51-62, Sep. 2018
    Detail / arXiv / GitHub

    Deep residual network with enhanced upscaling module for super-resolution
    J.-H. Kim, J.-S. Lee
    IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, pp. 913-921, Jun. 2018
    Detail / GitHub

    NTIRE 2018 challenge on single image super-resolution: methods and results
    R. Timofte et al. (including J.-H. Kim)
    IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, pp. 965-976, Jun. 2018
    Detail

    Assessing product design using photos and real products
    S.-E. Moon, J.-H. Kim, S.-W. Kim, J.-S. Lee
    ACM Conf. Human Factors in Computing Systems (CHI), Extended Abstract, pp. 1100-1107, May 2017
    Detail

    Travel photo album summarization based on aesthetic quality, interestingness, and memorableness
    J.-H. Kim, J.-S. Lee
    APSIPA Annual Summit and Conference, Dec. 2016
    Detail

    Awards

    Merit Academic Paper Award
    2020-2 Yonsei Superior Paper Awards, Dec. 2020

    2nd Place (Region 1)
    Yonsei-MCML team
    Super-Resolution Challenge on Perceptual Image Restoration and Manipulation (PIRM), in conjunction with ECCV, Sep. 2018
    Leaderboard / GitHub

    2nd Place (Region 2)
    Yonsei-MCML team
    Super-Resolution Challenge on Perceptual Image Restoration and Manipulation (PIRM), in conjunction with ECCV, Sep. 2018
    Leaderboard / GitHub