Research Interests

• Computer Vision and deep learning.

• 3D Pose Estimation/Mesh Reconstruction, Action recognition and Object detection.

Education

• M.S. at Artificial Intelligence Graduate School, Ulsan National Institute of Science and Technology (UNIST), Ulsan, South Korea.

  • Advisor: Seungryul Baek

  • Mar. 2022~

• B.S. at Dept. of Information and Communication Engineering, Inha university, Incheon, South Korea.

  • Mar. 2014~ Feb. 2022.

Publications

• Junsu Kim, Hoseong Cho, Jihyeon Kim, Yihalem Yimolal Tiruneh, Seungryul Baek, "SDDGR: Stable Diffusion-based Deep Generative Replay for Class Incremental Object Detection", in Proc. of IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), Seattle, USA, 2024. (​Highlight, 324/2719≈11.9% Accept ratio) 

• YunSeong Cho, Chanwoo Kim, Hoseong Cho, Yunhoe Ku, Eunseo Kim, Muhammadjon Boboev, Joonseok Lee, Seungryul Baek, "RMFER: Semi-supervised Contrastive Learning for Facial Expression Recognition with Reaction Mashup Video", in Proc. of IEEE Winter Conf. on Applications of Computer Vision (WACV), Hawaii, USA, 2024. (Oral presentation, 53/2,042≈2.6% Accept ratio) 

• Hoseong Cho, Chanwoo Kim, Jihyeon Kim, Seongyeong Lee, Elkhan Ismayilzada, Seungryul Baek, "Transformer-based Unified Recognition of Two Hands Manipulating Objects", in  Proc. of IEEE Conf. on Computer Vision and Pattern Recognition (CVPR), Vancouver, Canada, 2023.

• Hoseong Cho, Jihyeon Kim, Yunhoe Ku, Eunseo Kim, GeonU Kim and Seungryul Baek, "A Survey of Hand Pose Estimation", CKAIA, 2022.

• Jihyeon Kim, Junuk Cha, Yeongjun Choo, Minjae Lee, Jitae Yoo, Yeonghun Oh, Hoseong Cho, Eunseo Kim, Chungseok Choi, Honggyu Lim, Donguk Min, Eugene Shin, Seokun Kang and Seungryul Baek, "Free-form Masked Facial Image Inpainting via Exploiting 2D and 3D Information", CKAIA, 2022.

• Chanwoo Kim, Hoseong Cho, Hansol Lee, Yunseong Cho, Seungryul Baek, "Survey on Graph Attention Networks for Computer Vision Research", JKAIA, 2021

Projects

• Expanding Data Coverage through Incremental Learning and Development of Integrated Domain Recognition Systems, LG Electronics, 2023.1 - 2023.6 

• Research on high-speed hand pose estimation based on single RGB image, ETRI (Electronics and Telecommunications Research Institute), 2022.1~2022.12

• Lightweight hand pose model optimization

• Development of additional training algorithm for heatmap to improve overall 3D hand pose performance

• Generation of hand joint 3D data for hand poses suitable for AR glasses.

• Development of pre-processing software for action recognition datasets and development of dataset structuring and testing software, KETI (Korea Electronics Technology Institute), 2022.7~2023.12

  • • Construction of baseline model for action classification research.

    • Development of testing software for video classification.

Awards

• 1st place at HBHA challenge (Egocentric 3D hand pose estimation) held in ECCV 2022.

• 2nd place at HBHA challenge (Hand and object pose based action recognition) held in ECCV 2022.

Activity

• 3rd AI Novatus Academia (Computer Vision part) at UNIST AI Innovation Park, Ulsan, South Korea, 2022.

• Boost Camp AI Tech by Naver, Jan. 2021~ Jun. 2021.