Publications

I’m interested in the intersection of video and language. My research focuses on:

• Robust learning of information uniquely available in videos (e.g., temporal dynamics), minimizing language/image bias.
• Efficient modeling for video understanding: data efficiency (e.g., image-to-video adaptation) and computation efficiency (e.g., token merging/pruning, adaptive sampling).
• Actionable interpretability of video-language models, turning mechanistic insights into improved model behavior.

You can also find my papers on my Google Scholar profile.

Selected Publications

Map the Flow: Revealing Hidden Pathways of Information in VideoLLMs
Minji Kim*, Taekyung Kim*, Bohyung Han (*Equal Contribution)
ICLR 2026

Leveraging Temporal Contextualization for Video Action Recognition
Minji Kim, Dongyoon Han, Taekyung Kim, Bohyung Han
ECCV 2024

Towards Sequence-Level Training for Visual Tracking
Minji Kim*, Seungkwan Lee*, Jungseul Ok, Bohyung Han, Minsu Cho (*Equal Contribution)
ECCV 2022

Online Hybrid Lightweight Representations Learning: Its Application to Visual Tracking
Ilchae Jung, Minji Kim, Eunhyeok Park, Bohyung Han
IJCAI 2022

Other Projects

Learning Gaussian Models for Orientation-Aware Visual Tracking
Minji Kim, Bohyung Han
US Patent (App. 18/370,531)

Spatio-Temporal Modeling via Adaptive Frequency Filtering for Video Action Recognition
Minji Kim, Taehoon Kim, Jonghyeon Seon, Bohyung Han
Journal of Korean Institute of Information Scientists and Engineers (KIISE), 2024

Ensemble Modeling with Convolutional Neural Networks for Application in Visual Object Tracking
Minji Kim, Ilchae Jung, Bohyung Han
Journal of Korean Institute of Information Scientists and Engineers (KIISE), 2021

Top-down Thermal Tracking Based on Rotatable Elliptical Motion Model for Intelligent Livestock Breeding
Minji Kim, Wonjun Kim
Multimedia Systems, Springer, 2020