AI RESEARCH

TrackMAE: Video Representation Learning via Track Mask and Predict

arXiv CS.CV

ArXi:2603.27268v1 Announce Type: new Masked video modeling (MVM) has emerged as a simple and scalable self-supervised pre