MM-Motion: A Multimodal Human Motion Dataset Supporting Action Understanding and Injury Risk Evaluation

MM-Motion is a synchronized multimodal dataset integrating vision, IMU, and plantar pressure, providing a high-quality benchmark for action understanding and ankle sprain risk evaluation.

It is a multimodal dataset featuring 16 standardized actions performed by 33 subjects. The dataset comprises 933 IMU data files, 1,866 Kinect video files including over 260,000 RGB-D image pairs, 918 pressure insole data streams, and fundamental physiological metadata.

Raw data are available via separate access.