Min Ju Kim
Papers
2
Total Citations
19
H-Index
2
About
Min Ju Kim is a rising researcher in computer vision and robotics, specializing in action recognition and visual perception for autonomous systems. Their work focuses on developing efficient, occlusion-robust models that enable machines to understand human movements in real-world environments. Kim’s most-cited paper, “Fluxformer: Flow-Guided Duplex Attention Transformer via Spatio-Temporal Clustering for Action Recognition” (2023, 12 citations), introduces a novel vision transformer architecture that addresses the computational inefficiency of traditional transformers by using flow-guided attention and spatio-temporal clustering, achieving strong performance in robotics and automation tasks. In their second highly cited work, “Occluded Part-aware Graph Convolutional Networks for Skeleton-based Action Recognition” (2024, 7 citations), Kim tackles the critical challenge of occlusion—where body parts are hidden from view—by designing a part-aware graph convolutional network that maintains recognition accuracy even when visual data is incomplete. This contribution is vital for robots operating in cluttered or dynamic environments. With both papers published in top venues within just two years, Kim is establishing a reputation for advancing practical, real-time action recognition, bridging the gap between theoretical model design and deployment in autonomous systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2