About

Deokhyun Ahn 안덕현

PhD Candidate, Brain & Cognitive Engineering, Korea University · Student Researcher at KIST AI·Robotics

I am a Ph.D. candidate in Brain & Cognitive Engineering at Korea University and a student researcher at KIST AI·Robotics. My research focuses on video understanding, contextual dynamics, and efficient AI systems for social good.

Starting from sensing-based applications — including a SLAM-based smart helmet for firefighters and indoor drone trajectory tracking — I became interested in reliable real-world systems. I now study how video models can move beyond RGB and motion cues to capture contextual interactions, with recent focus on VLM-guided video anomaly detection for surveillance, especially target-domain transfer and efficient sampling.

Research keywords
Video Intelligence Video Understanding Video Anomaly Detection Video Action Recognition
01 News →
Dec 01, 2025
Received the AIR Outstanding Research Award from KIST AI·Robotics Institute, in recognition of research achievements during 2025.
Award
Feb 01, 2025
Received the Excellent Paper Award from Korea University Graduate School for outstanding research performance during doctoral studies.
Award
Nov 01, 2024
Received the Outstanding Oral Presentation Award at the KIST Academia–Research Convergence Conference for the talk “Multi-Aspect Contextual Dynamics for Anomaly Detection in Surveillance Videos.”
Talk
Jul 01, 2021
Received the Division Commander’s Commendation from the Capital Mechanized Infantry Division for a combat-development proposal on reconnaissance drone AI improvement and AI/big-data-based acoustic detection.
Service
02 Selected publications →
  1. CVPRW
    Look Closer: Action-Guided Dense Visual Dynamics for Proficiency Estimation
    Deokhyun Ahn, Hyukjin Kim, Jae-Ho Han, and 4 more authors
    In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW): Multimodal Human Motion Analysis, 2026
    Accepted (Pitch & Poster presentation)
  2. BMVC
    Learning Robust Representations for Few-Shot Action Recognition with Frame-Level Ambiguities
    Deokhyun Ahn, Yongjin Jo, Ui-Seok Lee, and 3 more authors
    In Proceedings of the British Machine Vision Conference (BMVC), 2026
    Under Review
  3. BMVC
    BUSTER: Adaptive Sampling for VLM-Guided Unsupervised Video Anomaly Detection
    Deokhyun Ahn, Yonghun Choi, Jae-Ho Han, and 2 more authors
    In Proceedings of the British Machine Vision Conference (BMVC), 2026
    Under Review
  4. TIP
    Where and What: Contextual Dynamics-Aware Anomaly Detection in Surveillance Videos
    Deokhyun Ahn, Yongjin Jo, Doyeon Kim, and 3 more authors
    IEEE Transactions on Image Processing, 2025
    IF 13.7 · top 2.2% in Electrical & Electronic Engineering
    Bib