This page showcases 3D object detection results on the ScanNet dataset. Each row compares the ground truth floor plan with results from three methods: V-DETR, SceneScript, and SpatialLM.
