News

On March 28th, we attended the KAIST Strawberry Party with some of our lab members.
It was a great time to catch up over delicious strawberries and refreshments.

Our laboratory alumnus, Senior Chiwan Song, visited our lab on March 8th and held a Hyundai Mobis information session. We were able to learn about the various activities at Hyundai Mobis and had the opportunity to ask questions about daily company life, which we had been curious about. Thanks to him, it was a very informative session.

Yoonki Cho has been the first recipient of the Research Impact Award in our lab.

The Research Impact Award honors students whose research has significantly impacted their academic community, as evidenced by the number of citations their papers have garnered.

Yoonki Cho’s paper, “Part-based pseudo label refinement for unsupervised person re-identification (Y Cho, WJ Kim, S Hong, SE Yoon),” was accepted for the IEEE/CVF Conference on Computer Vision and Pattern Recognition 2022. Also, it has been cited 120 times as of March 6, 2024. This indicates that the research has made a significant impact and contributed greatly to the field.

Congratulations to Yoonki!

Our lab members, Jeil and Taeguen, presented posters at the KRoC 2024 conference hosted by the Korean Robotics Society from February 21st to 24th, 2024. Jeil presented on the research topic of “Reinforcement Learning Based Quadrupedal Robot Controller for Adaptive Motion Generation”. Taeguen presented on the topic of “Manipulator-Assisted Navigation Among Movable Obstacles using Learned Robot-Obstacle Kinodynamics Model”.

Inkyu, Kyubeom, and Jumin graduated today. Congratulations on your graduation!

Wishing you a successful career ahead!

On January 19th, Professor Hwanjun Song from KAIST gave a seminar at our laboratory. He introduced his research area, “Learning with Noisy Labels.” His lecture was highly informative and especially beneficial for our lab’s research, which involves the use of various and large amounts of data.

Woojae Kim, a PhD students in our lab, was chosen as the winner of Qualcomm innovation fellowship Korea 2023. His finalist paper titled “Feature separation and Recalibaration for Adversarial Robustness” was selected as an highlight paper for CVPR 2023, and this achievement once again proves his excellence. Congraturations!

10월 27일에 계룡산 국립공원 수통골 방문하여 산책하고, 맛있는 점심식사를 하며 즐거운 시간을 보냈습니다. 오랜만에 가벼운 등산을 하며 맑은 공기를 마셨고, 연구실 사람들과 함께 이야기를 나누며 걷다보니 더욱 힐링되는 시간이 되었습니다. 맛있는 점심과 티타임을 즐기며, 친목도 더욱 돈독해졌습니다!

Prof. Yoon attended PG2023, held on Oct. between 10-13 in Daejeon, Korea. He acted as a session chair in the conference.

At the ICCV2023 conference, Guoyuan An presented his paper, “Towards Content-based Pixel Retrieval in Revisited Oxford and Paris.” This research introduced the innovative concept of pixel retrieval, along with associated benchmark datasets.

Xu Yin gave a presentation about his work at the ICCV2023 workshop (VIPriors). The paper is related to his journal paper titled “Contour-Aware Equipotential Learning for Semantic Segmentation“, published in TMM2022.

Prof. Yoon and his Ph.D. students — Jaeyoon Kim, Guoyuan An, and Xu Yin — attended the ICCV2023 conference, held in Paris, France from Oct. 2nd to 6th. At the conference, we had the pleasure of reconnecting with Yuchi Huo, an assistant professor at Zhejiang University and a former post-doctoral researcher at our lab.

우리 연구실에서 2014년도에 졸업하신 한국기술교육대학교 김덕수 교수님께서, “CUDA 기반 GPU 병렬 처리 프로그래밍” 저서를 발간하셨습니다. CUDA 기반 병렬처리 최적화를 배우고자하는 구독자들에게 추천합니다.

Our professor, Sung-Eui Yoon, gave an oral presentation at the KCCV2023 conference held between Aug 7 and 9 in Seoul, Korea. In this talk, he introduced “Feature Separation and Recalibration for Adversarial Robustness” which was chosen as a highlight paper (acceptance ratio: 2.6%) at CVPR2023.

Our paper titled “Confidence-Based Robot Navigation Under Sensor Occlusion with Deep Reinforcement Learning” was selected as one of the KAIST’s Research Highlights on the 2023 KAIST Annual R&D Report. This paper proposed a novel RL-based navigation method under sensor occlusion.

On July 21, Prof. Simon Woo, from Sungkyunkwan Univ., gave a seminar at our lab. He introduced his ongoing and published research on deep fake generation and detection topics.