Modular Action Concept Grounding in Semantic Video Prediction | IEEE Conference Publication | IEEE Xplore