Kangning Liu

Papers

Total Citations

H-Index

About

Kangning Liu is a researcher focused on advancing machine learning for fine-grained action identification, with particular emphasis on high-temporal-resolution analysis of human movement. Her major contributions center on addressing the limitations of existing action recognition systems, which typically identify coarse, long-duration activities. Liu's work pioneers the detection of sub-second actions, enabling more precise and responsive applications in robotics and smart health. She introduced the StrokeRehab benchmark dataset (2022, 5 citations), a critical resource for developing and evaluating models that identify rapid, subtle movements from video and kinematic data. Complementing this, her sequence-to-sequence modeling approach (2021, 2 citations) pushes the temporal resolution of action identification, allowing systems to distinguish between actions that occur in fractions of a second. These contributions are particularly impactful for rehabilitation monitoring, where capturing the precise timing and quality of patient movements is essential. Liu's research bridges the gap between coarse action recognition and the nuanced, real-time demands of clinical and robotic applications, establishing a foundation for more intelligent and responsive human-machine interaction systems.