Research map / 3D vision / 3D generation
Language Driven Occupancy Prediction
Zhu Yu, Bowen Pang, Lizhe Liu, Runmin Zhang, Qiang Li, Si-Yuan Cao, Maochun Luo, Mingxia Chen, Sheng Yang, Hui-Liang Shen
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use CLIP-aligned text labels from OV-Seg instead of image features as intermediates (ablation) | LOcc-BEVDet4D · Occ3D-nuScenes zero-shot ablation (256x704) | mIoU: 19.37 -> 22.95 | Improved |
| use CLIP-aligned text labels from OV-Seg instead of image features as intermediates (ablation) | LOcc-BEVFormer · Occ3D-nuScenes zero-shot ablation (256x704) | mIoU: 15.04 -> 18.62 | Improved |
| use multi-frame consecutive vocabularies instead of single-frame vocabularies (ablation) | LOcc-BEVDet4D · Occ3D-nuScenes zero-shot ablation (256x704) | mIoU: 22.15 -> 22.95 | Improved |
| use multi-frame consecutive vocabularies instead of single-frame vocabularies (ablation) | LOcc-BEVDet · Occ3D-nuScenes zero-shot ablation (256x704) | mIoU: 19.61 -> 20.29 | Improved |
| use multi-frame consecutive vocabularies instead of single-frame vocabularies (ablation) | pseudo-labeled GT · Occ3D-nuScenes zero-shot ablation (256x704) | mIoU: 22.66 -> 25.53 | Improved |
| use majority-voting voxelization instead of nearest-point voxelization (ablation) | LOcc-BEVDet4D · Occ3D-nuScenes zero-shot ablation (256x704) | mIoU: 22.68 -> 22.95 | Improved |
| use majority-voting voxelization instead of nearest-point voxelization (ablation) | pseudo-labeled GT · Occ3D-nuScenes zero-shot ablation (256x704) | mIoU: 23.80 -> 25.53 | Improved |
| use LiDAR point projection with occlusion instead of voxel-based model-view projection (ablation) | LOcc-BEVDet4D · Occ3D-nuScenes zero-shot ablation (256x704) | mIoU: 15.41 -> 22.95 | Improved |
| use LiDAR point projection with occlusion instead of voxel-based model-view projection (ablation) | LOcc-BEVDet · Occ3D-nuScenes zero-shot ablation (256x704) | mIoU: 14.49 -> 20.29 | Improved |
| swap OV-Seg model SAN for CAT-Seg (ablation) | LOcc-BEVDet4D · Occ3D-nuScenes zero-shot ablation (256x704) | mIoU: 22.68 -> 22.76 | No clear effect |
| swap OV-Seg model SAN for ODISE (ablation) | LOcc-BEVFormer · Occ3D-nuScenes zero-shot ablation (256x704) | mIoU: 18.62 -> 18.69 | No clear effect |
| add temporal input (BEVDet4D vs BEVDet) | LOcc · Occ3D-nuScenes zero-shot | mIoU: 20.29 -> 22.95 | Improved |
Closest papers by meaning
- OccMamba: Semantic Occupancy Prediction with State Space Models (CVPR 2025 · similarity 0.91)
- LOMA: Language-assisted Semantic Occupancy Network via Triplane Mamba (AAAI 2025 · similarity 0.89)
- OctOcc: High-Resolution 3D Occupancy Prediction with Octree (AAAI 2024 · similarity 0.89)
- LowRankOcc: Tensor Decomposition and Low-Rank Recovery for Vision-based 3D Semantic Occupancy Prediction (CVPR 2024 · similarity 0.89)
- 3D Occupancy Prediction with Low-Resolution Queries via Prototype-aware View Transformation (CVPR 2025 · similarity 0.88)
- Achieving Speed-Accuracy Balance in Vision-based 3D Occupancy Prediction via Geometric-Semantic Disentanglement (AAAI 2025 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
