Segtad: Precise temporal action detection via semantic segmentation

Publication
European Conference on Computer Vision