Research map / 3D vision / 3D generation
Test-Time 3D Occupancy Prediction
Fengyi Zhang, Xiangyu Sun, Huitong Yang, Zheng Zhang, Zi Huang, Yadan Luo
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 |
|---|---|---|---|
| add covariance-aware voxelization and sigmoid scale regulation (ablation) | TT-OccLiDAR · Occ3D-nuScenes 10% val | mIoU: 7.3 -> 18.3 | Improved |
| add covariance-aware voxelization and sigmoid scale regulation (ablation) | TT-OccCamera · Occ3D-nuScenes 10% val | mIoU: 4.2 -> 8.5 | Improved |
| inherit previous Gaussians (accumulate without tracking) (ablation) | TT-OccLiDAR · Occ3D-nuScenes 10% val | mIoU: 18.3 -> 23.5 | Improved |
| inherit previous Gaussians without tracking (ablation) | TT-OccLiDAR · Occ3D-nuScenes 10% val | pedestrian IoU: 25.5 -> 12.8 | Hurt |
| add dynamic Gaussian tracking (ablation) | TT-OccLiDAR · Occ3D-nuScenes 10% val | mIoU: 23.5 -> 25.6 | Improved |
| add dynamic Gaussian tracking (ablation) | TT-OccLiDAR · Occ3D-nuScenes 10% val | bus IoU: 9.6 -> 17.2 | Improved |
| add dynamic Gaussian tracking (ablation) | TT-OccCamera · Occ3D-nuScenes 10% val | mIoU: 14.1 -> 14.1 | No clear effect |
| use REX-Omni instead of OpenSeeD semantics (ablation) | TT-OccLiDAR · Occ3D-nuScenes | mIoU: 23.6 -> 27.4 | Improved |
| use GroundedSAM2 instead of OpenSeeD semantics (ablation) | TT-OccLiDAR · Occ3D-nuScenes | mIoU: 23.6 -> 21.3 | Hurt |
| use MapAnything instead of VGGT depth (ablation) | TT-OccCamera (REX-Omni) · Occ3D-nuScenes | mIoU: 15.9 -> 16.7 | Improved |
| use TT-OccLiDAR instead of SelfOcc (same OpenSeeD) | SelfOcc · Occ3D-nuScenes | RayIoU@4: 12.0 -> 25.8 | Improved |
| use TT-OccCamera instead of SelfOcc (same OpenSeeD) | SelfOcc · Occ3D-nuScenes | RayIoU@4: 12.0 -> 15.7 | Improved |
Closest papers by meaning
- ODG: Occupancy Prediction Using Dual Gaussians (NeurIPS 2025 · similarity 0.92)
- Achieving Speed-Accuracy Balance in Vision-based 3D Occupancy Prediction via Geometric-Semantic Disentanglement (AAAI 2025 · similarity 0.92)
- CSV-Occ: Fusing Multi-frame Alignment for Occupancy Prediction with Temporal Cross State Space Model and Central Voting Mechanism (ICML 2025 · similarity 0.91)
- OctOcc: High-Resolution 3D Occupancy Prediction with Octree (AAAI 2024 · similarity 0.91)
- SelfOcc: Self-Supervised Vision-Based 3D Occupancy Prediction (CVPR 2024 · similarity 0.91)
- OPUS: Occupancy Prediction Using a Sparse Set (NeurIPS 2024 · similarity 0.91)
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.
