MoVi stands for Large Multipurpose Motion and Video. It is a large multi-purpose motion video dataset that contains 20 predetermined daily and sports actions, and 1 self-selected action. These videos were completed by 60 women and 30 men. Three different hardware devices collected data five times, a total of 9 hours of motion capture data, 17 hours of video data from 4 angles (including a handheld camera), and 6.6 hours of IMU data.

This dataset can be used for research in the fields of action recognition and motion analysis, and based on it, researchers have also developed state-of-the-art skeletal motion estimation methods.