SPACE

3DEnvironmentIntroduced 2021-08-13

SPACE is a simulator for physical Interactions and causal learning in 3D environments. The SPACE simulator is used to generate the SPACE dataset, a synthetic video dataset in a 3D environment, to systematically evaluate physics-based models on a range of physical causal reasoning tasks. Inspired by daily object interactions, the SPACE dataset comprises videos depicting three types of physical events: containment, stability and contact.

Related Benchmarks

SPACE (Opinion Summarization)/Text Summarization/ROUGE-2 SPACE (Opinion Summarization)/Text Summarization/ROUGE-L SpaceNet 1/10-shot image generation/Mean IoU SpaceNet 1/16k/F1 Score SpaceNet 1/2D Classification/F1 Score SpaceNet 1/2D Object Detection/F1 Score SpaceNet 1/3D/F1 Score SpaceNet 1/Object Detection/F1 Score SpaceNet 1/Semantic Segmentation/Mean IoU SpaceNet 2/16k/F1 Score (Avg. over Cities)SpaceNet 2/2D Classification/F1 Score (Avg. over Cities)SpaceNet 2/2D Object Detection/F1 Score (Avg. over Cities)SpaceNet 2/3D/F1 Score (Avg. over Cities)SpaceNet 2/Object Detection/F1 Score (Avg. over Cities)