VividNet

The dataset used in the paper is a scene-graph, which represents the semantic information needed to perform intuitive physics.

BibTex: