Something-else: Compositional action recognition with spatial-temporal interaction networksPublished in Computer Vision and Pattern Recognition, 2020Download paper hereShare on Twitter Facebook LinkedIn Previous Next