SDVScenes

A novel large-scale dataset for spatio-temporal tagging of self-driving scenes, containing 15 attributes relating to vehicle and pedestrian density, the actions of each actor, the speed of each actor, interactions between actors, and the topology of the road map.