RenderOcc: Vision-Centric 3D Occupancy Prediction with 2D Rendering Supervision

arXiv:2309.09502v2 Announce Type: replace
Abstract: 3D occupancy prediction holds significant promise in the fields of robot perception and autonomous driving, which quantifies 3D scenes into grid cells with semantic labels. Recent works mainly utilize complete occupancy labels in 3D voxel space for supervision. However, the expensive annotation process and sometimes ambiguous labels have severely constrained the usability and scalability of 3D occupancy models. To address this, we present RenderOcc, a novel paradigm for training 3D occupancy models only using 2D …

3d scenes abstract annotation arxiv autonomous autonomous driving cells cs.cv driving fields grid labels perception prediction process rendering robot robot perception semantic space supervision type vision voxel