Ho-Deok Jang

Research Engineer

Video Team, CLOVA, NAVER Corp.

jhodeok@gmail.com

[CV] [github]

I'm a Research Engineer at NAVER Clova. My current research interest includes following tasks:

  • Visual recognition on image/video

    • Object detection, Instance/panoptic segmentation, Person keypoints estimation.

  • 3D vision

    • Mono/stereo depth estimation, Neural scene representation (e.g., NeRF)

Publication

Self-Supervised Monocular Depth Estimation with Isometric-Self-Sample-Based Learning

Geonho Cha, Ho-Deok Jang, Dongyoon Wee

IEEE Robotics and Automation Letters (RA-L), 2022.

[pdf]

Propose-and-Attend Single Shot Detector

Ho-Deok Jang, Sanghyun Woo, Philipp Benz, Jinsun Park, In So Kweon

IEEE Winter Conference on Applications of Computer Vision (WACV), 2020. (Oral)

[pdf], [video oral], [video demo]

Fast Perception, Planning, and Execution for a Robotic Butler: Wheeled Humanoid M-Hubo

Moonyoung Lee, Yujin Heo, Jinyong Park, Hyundae Yang, Ho-Deok Jang, Philipp Benz, Hyunsub Park, In So Kweon, Jun-Ho Oh

IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2019.

[pdf], [video]

Gated Bidirectional Feature Pyramid Network for Accurate One-shot Detection

Sanghyun Woo, Soonmin Hwang, Ho-Deok Jang, In So Kweon

Machine Vision and Applications (MVA), 2019.

[pdf]

Project

Visual AI system for NAVER map

  • Member : NAVER CLOVA Video Team (w/ Min-Young Chang as intern), GLACE Team

  • Task : Developing a visual recognition system for parsing road view information under purpose of enhancing NAVER map services.

  • Contribution: object detection/tracking dev codebase, modeling of object detector w/ class imbalance, modeling of lane detector

  • Software : Python, PyTorch

  • Link : [video]

Visual recognition algorithm for robot manipulation

  • Member : Ho-Deok Jang and Philipp Benz

  • Task : Developing a visual recognition system for robot manipulation. Specifically, the work is to design a 2d/3d object detection algorithms based on an RGB-D sensor and build a custom dataset for the task. Refer to a demo video below for the results.

  • Advisor : Prof. Jun-Ho Oh and Prof.In So Kweon

  • Software : Python, PyTorch, ROS

  • Link: [video]

Paper Reproduce

Implemented and reproduced "Zhu et al. Feature Selective Anchor-Free Module for Single-Shot Object Detection. CVPR, 2019." in PyTorch (Sep. 2019).

  • Task: object detection

  • Note: To my best knowledge, this is the "first" open source code that implements and reproduces this paper on github at the time of upload.

[code]

Implemented and reproduced "Perez et al. FiLM: Visual Reasoning with a General Conditioning Layer. AAAI, 2018." in PyTorch (Sep. 2019).

  • Task: visual question answering (VQA)

[code]

Implemented and reproduced "Santoro et al. A simple neural network module for relational reasoning. NIPS, 2017. " in PyTorch (Sep. 2019).

  • Task: visual question answering (VQA)

[code]