Research map / 3D vision / 3D generation
Tri-Perspective View for Vision-Based 3D Semantic Occupancy Prediction
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 three orthogonal TPV planes instead of a single BEV plane | Vision-based LiDAR segmentation · nuScenes validation | mIoU: 56.2 -> 68.9 | Helped |
| Use TPVFormer instead of MonoScene | Vision-based semantic scene completion · SemanticKITTI test | SSC mIoU: 11.08 -> 11.26 | Helped |
| Use TPVFormer instead of MonoScene | Vision-based semantic scene completion · SemanticKITTI validation | SSC mIoU: 11.08 -> 11.36 | Helped |
| Use TPVFormer instead of JS3C-Net | Vision-based semantic scene completion · SemanticKITTI validation | SC IoU: 38.98 -> 35.61 | Hurt |
| Apply CE to voxel predictions and Lovasz to point predictions (ablation) | TPVFormer LiDAR segmentation training · nuScenes validation | Point mIoU: 50.66 -> 64.80 | Helped |
| Use point predictions for both CE and Lovasz supervision (ablation) | TPVFormer LiDAR segmentation training · nuScenes validation | Voxel mIoU: 63.17 -> 49.94 | Hurt |
| Increase TPV resolution from 100x100x8 to 200x200x16 while reducing feature dimension from 256 to 128 (ablation) | TPVFormer · nuScenes LiDAR segmentation validation | Point mIoU: 64.15 -> 68.86 | Helped |
| Increase resolution from 100x100 to 200x200 (ablation) | BEVFormer · nuScenes LiDAR segmentation validation | Point mIoU: 50.37 -> 56.21 | Helped |
| Increase HCAB blocks from 2 to 4, with total attention modules fixed (ablation) | TPVFormer · SemanticKITTI semantic scene completion validation | SC IoU: 35.55 -> 35.79 | Helped |
| Use a moderate HCAB/HAB balance of 3/2 rather than 4/0 (ablation) | TPVFormer · SemanticKITTI semantic scene completion validation | SSC mIoU: 10.82 -> 11.36 | Helped |
| Remove HAB blocks, using 4 HCAB and 0 HAB instead of 3 HCAB and 2 HAB (ablation) | TPVFormer · SemanticKITTI semantic scene completion validation | SSC mIoU: 11.36 -> 10.82 | Hurt |
Closest papers by meaning
- COTR: Compact Occupancy TRansformer for Vision-based 3D Occupancy Prediction (CVPR 2024 · similarity 0.91)
- SDGOCC: Semantic and Depth-Guided Bird's-Eye View Transformation for 3D Multimodal Occupancy Prediction (CVPR 2025 · similarity 0.89)
- Achieving Speed-Accuracy Balance in Vision-based 3D Occupancy Prediction via Geometric-Semantic Disentanglement (AAAI 2025 · similarity 0.89)
- CSV-Occ: Fusing Multi-frame Alignment for Occupancy Prediction with Temporal Cross State Space Model and Central Voting Mechanism (ICML 2025 · similarity 0.88)
- Tri-Perspective View Decomposition for Geometry-Aware Depth Completion (CVPR 2024 · similarity 0.88)
- SOGDet: Semantic-Occupancy Guided Multi-View 3D Object Detection (AAAI 2024 · similarity 0.88)
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.