Learning Semantic Relationships for Better Action Retrieval in Images

doi 10.1109/cvpr.2015.7298713