Hyunjoon Lee

Working at Kakao Brain

Research interests

  • Computer vision using deep learning - object detection; object tracking; geometric scene understanding; face detection; pose estimation

  • Image enhancement and manipulation

  • Image based camera calibration and geometric scene understanding

Experience

  • (Sep 2020 ~ present) Computer vision research engineer at Kakao Brain

  • (Jul 2015 ~ Aug 2020) Computer vision software engineer at Intel Korea

  • (Sep 2014 ~ Jun 2015) Post Doc in CSE, POSTECH

  • (Sep 2005 ~ Aug 2014) Ph.D in CSE, POSTECH

  • (Jun 2012 ~ Sep 2012) Tech transfer intern at Adobe

  • (Mar 2001 ~ Aug 2005) B.S. in CSE, POSTECH (Magna Cum Laude)

Projects

Object tracking using conventional methods and deep learning:

  • Implemented and improved state-of-the-art object tracking algorithms in terms of speed and quality.

  • Developed product quality algorithms which can be used in edge devices.

Deep learning based face and object detection:

  • Developed lightweight face and object detectors, including key-point detection.

Development of sparse and quantized networks:

  • Developed and trained models for Intel DL platforms.

Upright photo:

  • Researched on core algorithms of upright photo, including robust single image camera calibration and automatic adjustment system.

  • Worked as a tech transfer intern at Adobe and developed Upright for Adobe Lightroom.

Awards

  • ACCV'12 Special Session on RGB-D Camera Application Competition, Best Kinect-Application Award, 2012.

  • POSTECH Programming Contest, Honor Prize, 2005.

Publications

  • Yeonkyeong Lee, Taeho Choi, Hyunsung Go, Hyunjoon Lee, Sunghyun Cho, and Junho Kim. Exp-GAN: 3D-Aware Facial Image Generation with Expression Control. In Proc. ACCV, accepted for publication.

  • Jinwoo Lee, Hyunsung Go, Hyunjoon Lee, Sunghyun Cho, Minhyuk Sung, and Junho Kim. CTRL-C: Camera calibration TRansformer with Line-Classification. In Proc. ICCV, 2021.

  • Jinwoo Lee, Minhyuk Sung, Hyunjoon Lee, and Junho Kim. Neural Geometric Parser for Single Image Camera Calibration. In Proc. ECCV, 2020.

  • Junho Jeon, Hyunjoon Lee, Henry Kang, and Seungyong Lee. Scale-aware Structure-Preserving Texture Filtering. Computer Graphics Forum (Pacific Graphics 2016), 2016.

  • Hyunjoon Lee, Junho Jeon, Junho Kim, and Seungyong Lee. Structure-Texture Decomposition of Images with Interval Gradient. Computer Graphics Forum, 2016. doi:10.1111/cgf.12875

  • Hojin Cho, Hyunjoon Lee, and Seungyong Lee. Radial bright channel prior for single image vignetting correction. In Proc. ECCV, pages 189–202, 2014.

  • Hojin Cho, Hyunjoon Lee, Henry Kang, and Seungyong Lee. Bilateral texture filtering. ACM TOG, 33:128, 2014.

  • Hyunjoon Lee, Eli Shechtman, Jue Wang, and Seungyong Lee. Automatic upright adjustment of photographs with robust camera calibration. IEEE PAMI, 36:833–844, 2014.

  • Junho Jeon, Daehoon Yoo, Chiyoung Lee, Hyunjoon Lee, Junho Kim, and Seungyong Lee. Kinect Reality: Real-time 3D Scanning Based Augmented Reality. In Proc. ACCV Workshop on RGB-D Camera, 2012.

  • Hyunjoon Lee, Eli Shechtman, Jue Wang, and Seungyong Lee. Automatic upright adjustment of photographs. In Proc. CVPR, pages 877–884, 2012.

  • Hyunjoon Lee, Minsu Ahn, and Seungyong Lee. Displaced subdivision surfaces of animated meshes. Computers and Graphics, 35:532–541, 2011.

  • Sunghyun Cho, Hyunjoon Lee, and Seungyong Lee. Image decomposition using deconvolution. In Proc. ICIP, pages 1965–1968, 2010.

  • Hyunjoon Lee, Sungtae Kwon, and Seungyong Lee. Real-time pencil rendering, In Proc. NPAR, pages 37–45, 2006.

Patents

  • Byungseok Roh, Minje Park, Hyunjoon Lee, Seok-Yong Byun. Partial inference path in general object detection networks for efficient video processing, Patent-pending, 2019

  • Seungyong Lee, Hojin Cho, and Hyunjoon Lee. Method of Vignetting Correction Using Radial Bright Channel Prior, Recording Medium of performing the same, Apparatus of performing the same, KR Patent applied, 2014.

  • Seungyong Lee, Hyunjoon Lee, and Hojin Cho. Method of Texture Filtering Using Patch Shift, Recording Medium of performing the same, Apparatus of performing the same, KR Patent applied, 2014.

  • Elya Shechtman, Jue Wang, Hyunjoon Lee, and Seungyong Lee. Camera Calibration and Automatic Adjustment of Images, US Patent applied, 2013.

  • Elya Shechtman, Jue Wang, Hyunjoon Lee, and Seungyong Lee. Automatic Adjustment of Images using a Homography, US Patent applied, 2013.

Programming languages and tools

  • Languages: C/C++, Python

  • Tools: Matlab