DoMSEV Multimodal Video Dataset

DoMSEV stands for Dataset of Multimodal Semantic Egocentric Video, which is a multimodal video dataset based on personal activities. It contains 80 hours of RGB-D data, IMU data, and GPS data. The videos are annotated with the following: recorder profile, frame scene, activities, interaction, and attention. This dataset can be used to study the problem of fast-forwarding videos smoothly without losing relevant content.