A brief overview of HiST-VQ, an unsupervised AI method that automatically segments continuous skeleton motion into meaningful human actions. The video explains the core challenge, the hierarchical subaction-to-action approach, and the model’s state-of-the-art results without labels or timestamps.
