HunminYang.jpeg

I work as a senior AI researcher at ADD. Concurrently, I’m a PhD candidate at KAIST and a member of Visual Intelligence Lab., advised by Prof. Kuk-Jin Yoon.

My research aims to develop trustworthy visual intelligence which assists decision-making processes regarding real-world problems. To do this, I research in the interdisciplinary fields of computer vision and machine learning. My major interests currently lie in robust model training, domain generalization/adaptation, and adversarial robustness.

  • Keywords: Computer Vision, Machine Learning, Trustworthy AI

News

Aug 12, 2024 :trophy: PDCL-Attack selected as oral presentation at ECCV 2024.
Jul 1, 2024 📝 One paper accepted to ECCV 2024.
Apr 29, 2024 📝 One paper accepted to CVPRW 2024.
Dec 9, 2023 📝 One paper accepted to AAAI 2024.
Aug 9, 2023 📝 One paper accepted to ICCVW 2023.
Jul 18, 2023 📝 Two papers accepted to ICCV 2023.
Mar 3, 2022 📝 One paper accepted to CVPR 2022.
Dec 3, 2021 :trophy: I won the National Grant for Defense Research and Development from DAPA.

Experience

Jan, 2020 - Present Agency For Defense Development (ADD), Daejeon, South Korea
Senior AI Researcher
  • Synthetic data generation framework for training DL models (D-GEN)
  • Small & occluded object detection with CG-based synthetic training images
  • Adversarial attack & defense for robust AI
May, 2017 - Dec, 2019 Agency For Defense Development (ADD), Daejeon, South Korea
AI Researcher
  • Big data platform for large-scale intelligent video analytics (D-NET)
  • Accelerating the distributed deep learning inference on multi-GPU with Hadoop-Spark
  • Hosting a data science competition for satellite image recognition with AI
Feb, 2014 - May, 2017 Agency For Defense Development (ADD), Daejeon, South Korea
Specialized Research Staff (Military Service)

Education

Sep, 2021 - Present Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea
PhD in Mechanical Engineering
  • Research Area: Computer Vision & Machine Learning
  • Advisor: Prof. Kuk-Jin Yoon
Feb, 2012 - Feb, 2014 Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea
MS in Mechanical Engineering
  • Research Area: 3D Sound Perception
  • Advisor: Prof. Youngjin Park
Feb, 2011 - Aug, 2011 Royal Melbourne Institute of Technology (RMIT), Melbourne, Australia
Exchange Student (High Distinction)
Feb, 2007 - Feb, 2012 Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea
BS in Mechanical Engineering (Magna Cum Laude)

Selected Publications

  1. Prompt-Driven Contrastive Learning for Transferable Adversarial Attacks
    Hunmin Yang, Jongoh Jeong, and Kuk-Jin Yoon
    European Conference on Computer Vision (ECCV), 2024
    Oral Presentation (< 2.3%)
    Paper Project Page
  2. Object-Centric Domain Randomization for 3D Shape Reconstruction in the Wild
    Junhyeong Cho, Kim Youwang,  Hunmin Yang, and Tae-Hyun Oh
    arXiv preprint, 2024
    CVPR Workshop on Foundation Models (WFM), 2024
    Paper Project Page
  3. FACL-Attack: Frequency-Aware Contrastive Learning for Transferable Adversarial Attacks
    Hunmin Yang*, Jongoh Jeong*, and Kuk-Jin Yoon (*equal contribution)
    Association for the Advancement of Artificial Intelligence (AAAI), 2024
    ICCV Workshop on Adversarial Robustness In the Real World (AROW), 2023
    Paper Project Page
  4. PromptStyler: Prompt-driven Style Generation for Source-free Domain Generalization
    Junhyeong Cho, Gilhyun Nam, Sungyeon Kim,  Hunmin Yang, and Suha Kwak
    IEEE/CVF International Conference on Computer Vision (ICCV), 2023
    Paper Project Page
  5. ACTIVE: Towards Highly Transferable 3D Physical Camouflage for Universal and Robust Vehicle Evasion
    Naufal Suryanto, Yongsu Kim, Harashta Tatimma Larasati, Hyoeun Kang, Thi-Thu-Huong Le, Yoonyoung Hong,  Hunmin Yang, Se-Yoon Oh, and Howon Kim
    IEEE/CVF International Conference on Computer Vision (ICCV), 2023
    Paper Project Page
  6. DTA: Physical Camouflage Attacks Using Differentiable Transformation Network
    Naufal Suryanto, Yongsu Kim, Hyoeun Kang, Harashta Tatimma Larasati, Youngyeo Yun, Thi-Thu-Huong Le,  Hunmin Yang, Se-Yoon Oh, and Howon Kim
    IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022
    Paper Project Page

Professional Services

Academic Reviewer
  • IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023, 2024
  • IEEE/CVF International Conference on Computer Vision (ICCV), 2023
  • European Conference on Computer Vision (ECCV), 2024
Technology Transfer
  • Synthetic Data Generation Techniques for Training DL models (SI Analytics, JCORP SYSTEM, JEIOS, Xiilab), 2020, 2021, 2022, 2023
  • Camouflage Pattern Generation Techniques for Attacking DL models (SmartM2M), 2022
  • Big Data Platform Techniques for Real-time Video Recognition (Xiilab), 2019

Honors and Awards

  • Selected as Oral Presentation (< 2.3%), ECCV, 2024
    • Prompt-Driven Contrastive Learning for Transferable Adversarial Attacks
  • National Grant for Defense Research and Development, DAPA, 2021
    • Synthetic Data Generation for Defense AI
  • Defense Science Award, ADD, 2019
    • Improving Distributed Multi-GPU Computing for Large-scale Intelligent Video Analytics
  • High Achievement Award, ADD, 2018
    • Big Data Platform Development and Synthetic Data Generation
  • Excellent Paper Award, KSNVE, 2013
    • Sweet Spot Analysis of Linear Array System with a Large Number of Loudspeakers by Geometrical Approach Method