Curriculum Vitae
WORK EXPERIENCE
Visual Display Business in Samsung Electronics, (Mar. 2021 – Present)
(화질알고리즘 파트 in H/W Platform Lab)
Frame Rate Conversion (FRC) Team
화질 관련 TV SoC 개발 (H/W, NPU, DSP)
Deep learning-based object detection
Deep learning-based optical flow estimation, video frame interpolation
EDUCATION
Ph.D. in Electrical Engineering, KAIST (Mar. 2017 - Feb. 2021)
GPA : 4.21 / 4.30
Advisor: Prof. Munchurl Kim (http://viclab.kaist.ac.kr)
Thesis: "Deep Predictive Video Compression using Mode-Selective Uni- and Bi-directional Predictions based on Multi-frame Hypothesis"
M.S. in Electrical Engineering, KAIST (Mar. 2015 - Feb. 2017)
GPA : 4.22 / 4.30
Advisor: Prof. Munchurl Kim (http://viclab.kaist.ac.kr)
Thesis : "A Study on Learning-based Linear and Nonlinear Sparsifying Transforms for Image Compression"
B.S. in Electrical Engineering, KAIST (Feb. 2011 - Feb. 2015)
GPA : 4.07 / 4.30
RESEARCH INTERESTS
Deep learning-based Frame interpolation/Optical flow estimation
Deep Image/Video Compression
Deep learning-based compression artifact reduction
Low-level visions (ex. Image/Video Super-Resolution, Image/Video denoising)
High-level visions (ex. Deep learning-based Object detection)
PUBLICATIONS
INTERNATIONAL
Woonsung Park and Munchurl Kim. "Deep Predictive Video Compression using Mode-Selective Uni-and Bi-directional Predictions based on Multi-frame Hypothesis," IEEE ACCESS, Dec. 2020.
Citations: 7 (updated in 2023-07-27)
S. Son et al. (including Woonsung Park), "AIM 2020 Challenge on Video Temporal Super-Resolution," European Conference on Computer Vision Workshop (ECCVW), Aug. 2020.
Citations: 19 (updated in 2023-07-27)
Woonsung Park, Sehwan Ki, and Munchurl Kim, "The factsheet of Deep Video Compression based on the Context-Adaptive Entropy Model for P-frame Compression," IEEE Conference on Computer Vision and Pattern Recognition Workshop (CVPRW), 2020.
Woonsung Park, Bumshik Lee and Munchurl Kim, "Fast Computation of Integer DCT-V, DCT-VIII and DST-VII for Video Coding," IEEE Transactions on Image Processing, vol. 28, issue 12, pp. 5839 - 5851, Dec. 2019.
Citations: 22 (updated in 2023-07-27)
Woonsung Park and Munchurl Kim. "Deep predictive video compression with bi-directional prediction," arXiv preprint arXiv:1904.02909 (2019).
Citations: 4 (updated in 2023-07-27)
S. Nah et al. (including Woonsung Park), "AIM 2019 Challenge on Video Temporal Super-Resolution: Methods and Results," 2019 IEEE/CVF International Conference on Computer Vision Workshop (ICCVW), Seoul, Korea (South), 2019, pp. 3388-3398.
Citations: 37 (updated in 2023-07-27)
URL: https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9022180&isnumber=9021948 (doi: 10.1109/ICCVW.2019.00421)
Woonsung Park, Munchurl Kim, "CNN-based In-loop Filtering for Coding Efficiency Improvement," IEEE Image Video and Multidimensional Signal Processing (IVMSP) workshop, Bordeaux, France, 11-12 July, 2016.
Citations: 200 (updated in 2023-07-27)
DOMESTIC
1. 박운성, 김문철, "DenseNet 기반의 이미지 압축 ," 한국방송·미디어공학회, 한국방송미디어공학회 하계학술발표대회 논문집, 2018.6.20-22, pp. 272-275,제주도 한라대학교.
2. 박운성, 이범식, 김문철, "비디오 압축을 위한 Winograd FFT 기반 DCT-5 저복잡도 연산 기법," 한국통신학회 동계종합학술발표회 논문집, 2018.1.17-19, pp. 286-287, 강원도 하이원 리조트.
3. 박운성, 이범식, 김문철, "DST-7과 DCT-8의 저복잡도 연산 알고리즘," 대한전자공학회, 2017년도 대한전자공학회 하계종합학술대회 논문집, pp. 691-694, 2017. 6. 29-7.1, 해운대 그랜드 호텔.
4. 박운성, 이범식, 김문철, 구문모, "DCT-5의 저복잡도 연산 알고리듬 설계," 한국통신학회, 2017년 한국통신학회 춘계 프로그램 논문집, pp. 3-4, 2017. 6. 9, 조선대학교 전자정보공과대학.
5. 박운성, 김문철, "컨볼루션 신경망을 이용한 고효율 비디오 부호화에서의 인-루프 필터," 한국방송·미디어공학회 하계학술대회, 2016년 6월 29일 - 7월 1일, 제주한라대학교 금호세계교육관.
AWARDS AND HONORS
2nd Place Award, Video Temporal Super-Resolution Challenge, Advances in Image Manipulation workshop and challenges on image and video manipulation (AIM) 2020, ECCV 2020. [certificate]
Best paper Award, "Low Complexity Computation of DCT-V for Video Compression", Spring Conference of The Korean Institute of Communications and Information Sciences, June 2017.
PROJECTS
National Research Foundation of Korea, Deep learning based coding artifact reduction, Mar. 2016 – Feb. 2017.
LG Electronics, Low Complexity Computation of Transform Kernels in Video Compression, Mar. 2017 - Feb. 2018.
National Research Foundation of Korea, Deep learning based Video Compression, Mar. 2018 – Feb. 2021.
National Research Foundation of Korea, Deep learning based Frame Interpolation, Mar. 2018 – Feb. 2021.
INVITED TALKS
Samsung Electronics, Recent Trends in deep learning based frame interpolation (Nov. 2018)
SKILLS AND TECHNIQUES
Programming Languages - C, MATLAB, Python
Deep learning framework - Tensorflow, Pytorch
CONTACT
pys53091@gmail.com