Bumsoo Kim

Assistant Professor at Chung-Ang University

School of Computer Science and Engineering

bumsoo [at] cau [dot] ac [dot] kr
Google Scholar / Github / LinkedIn

About

I am an assistant professor at the School of Computer Science and Engineering in Chung-Ang University (CAU). My research area covers building efficient pipelines for Multi-modal representation learning and Scene Understanding. During my PhD, I was fortunate to be advised by Jaewoo Kang and Hyunwoo J. Kim. Previously I worked 3 years at LG AI Research and 2 years at KakaoBrain as a research scientist.

Education

• Ph.D. Computer Science and Engineering, Korea University (Mar 2016 ~ Feb 2022)
  • Full Scholarship (Anam scholarship)
  • Thesis: Parallel Architectures Towards Real-Time Human-Object Interaction Detection
    •
• B.S. Computer Science and Engineering, Korea University (Mar 2012 ~ Feb 2016)
  • Full Scholarship (Anam scholarship)
  • Entered as Top of the Department
    •

Selected Publications

Bi-directional Contextual Attention for 3D Dense Captioning
Minjung Kim, Hyung Suk Lim, Soonyoung Lee, Bumsoo Kim†, Gunhee Kim†
In ECCV 2024, Oral Presentation
[Paper] [arXiv] [Code] [Video]

See It All: Contextualized Late Aggregation for 3D Dense Captioning
Minjung Kim, Hyung Suk Lim, Seung Hwan Kim, Soonyoung Lee, Bumsoo Kim†, Gunhee Kim†
In ACL 2024 Findings
[Paper] [arXiv] [Code] [Video]

Expediting Contrastive Language-Image Pretraining via Self-distilled Encoders
Bumsoo Kim, Jinhyung Kim, Yeonsik Jo, Seung Hwan Kim
In AAAI 2024
[Paper] [arXiv] [Code] [Video]

Uncertainty-Guided SpatioTemporal Transformer for 3D Human Pose and Shape Estimation on Videos
Minsoo Lee*, Hyunmin Lee*, Bumsoo Kim†, Seung Hwan Kim†
In WACV 2024, Oral Presentation (Top 2.5%)
[Paper] [arXiv] [Code] [Video]

Misalign, Contrast then Distill: Rethinking Misalignments in Language-Image Pretraining
Bumsoo Kim, Yeonsik Jo, Jinhyung Kim, Seung Hwan Kim
In ICCV 2023
[Paper] [arXiv] [Code] [Video]

UniCLIP: Unified Framework for Contrastive Language-Image Pre-training
Janghyeon Lee*, Jongsuk Kim*, Hyounguk Shon, Bumsoo Kim, Seung Hwan Kim, Honglak Lee, Junmo Kim
In NeurIPS 2022
[Paper] [arXiv] [Code] [Video]

MSTR: Multi-Scale Transformer for End-to-End Human-Object Interaction Detection
Bumsoo Kim, Jonghwan Mun, Kyoung-Woon On, Minchul Shin, Junhyun Lee, Eun-Sol Kim
In CVPR 2022
[Paper] [arXiv] [Code] [Video]

HOTR: End-to-End Human-Object Interaction Detection with Transformers
Bumsoo Kim, Junhyun Lee, Jaewoo Kang, Eun-Sol Kim, Hyunwoo J. Kim
In CVPR 2021, Oral Presentation (Top 4%)
[Paper] [arXiv] [Code] [Video]

UnionDet: Union-Level Detector Towards Real-Time Human-Object Interaction Detection
Bumsoo Kim*, Taeho Choi*, Jaewoo Kang, Hyunwoo J. Kim
In ECCV 2020
[Paper] [arXiv] [Code] [Video]

News

2024.07 I will be joining Chung-Ang University from 2024.09 as an Assistant Professor!
2024.07 1 paper (BiCA) accepted to ECCV 2024 as Oral Presentation!
2024.05 1 paper (SIA) accepted to ACL 2024 Findings.

2023.12 1 paper (ECLIPSE) accepted to AAAI 2024.
2023.10 1 paper (UNSPAT) accepted to WACV 2024 as Oral Presentation!
2023.07 1 paper (MCD) accepted to ICCV 2023.

2022.09 1 paper (UniCLIP) accepted to NeurIPS 2022.
2022.03 Started working as a Research Scientist in Vision Lab at LG AI Research.
2022.02 1 paper (MSTR) accepted to CVPR 2022.

2021.02 1 paper (HOTR) accepted to CVPR 2021 as Oral Presentation!

2020.11 Started working as a Research Scientist in Video Intelligence Team at KakaoBrain.
2020.08 Started Internship at KakaoBrain (Host: Eun-Sol Kim).
2020.07 1 paper (UnionDet) accepted to ECCV 2020.