Research map / 3D vision / 3D generation
Semantic Causality-Aware Vision-Based 3D Occupancy Prediction
Dubing Chen, Huan Zheng, Yucheng Zhou, Xianfei Li, Wenlong Liao, Tao He, Pai Peng, Jianbing 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 |
|---|---|---|---|
| remove depth supervision loss (ablation) | BEVDetOcc · Occ3D-nuScenes ablation (BEVDetOcc) | mIoU: 37.1 -> 36.8 | Hurt |
| add causal loss (ablation) | BEVDetOcc · Occ3D-nuScenes ablation (BEVDetOcc) | mIoU: 37.1 -> 37.6 | Improved |
| use unbiased estimator for causal loss (ablation) | BEVDetOcc · Occ3D-nuScenes ablation (BEVDetOcc) | mIoU: 37.6 -> 37.5 | No clear effect |
| remove post-lifting convolutions (ablation) | BEVDetOcc · Occ3D-nuScenes ablation (BEVDetOcc) | mIoU: 37.1 -> 37.3 | No clear effect |
| add channel-grouped lifting (ablation) | BEVDetOcc · Occ3D-nuScenes ablation (BEVDetOcc) | mIoU: 37.3 -> 37.6 | Improved |
| add learnable camera offsets (with soft filling) (ablation) | BEVDetOcc · Occ3D-nuScenes ablation (BEVDetOcc) | mIoU: 37.6 -> 37.9 | Improved |
| add normalized convolution (ablation) | BEVDetOcc · Occ3D-nuScenes ablation (BEVDetOcc) | mIoU: 37.9 -> 38.3 | Improved |
| add SCAT and causal loss | BEVDetOcc · Occ3D-nuScenes (single frame) | mIoU: 37.1 -> 38.3 | Improved |
| add SCAT and causal loss | ALOcc · Occ3D-nuScenes (single frame) | mIoU: 40.1 -> 40.9 | Improved |
| add SCAT and causal loss under camera noise | BEVDetOcc · Occ3D-nuScenes (Gaussian noise 0.1 on camera params) | mIoU relative drop: -32.3% -> -7.3% | Improved |
| add SCAT and causal loss under camera noise | ALOcc · Occ3D-nuScenes (Gaussian noise 0.1 on camera params) | mIoU relative drop: -21.9% -> -3.3% | Improved |
| replace depth-based LSS with semantic-causal-locality-aware LSS | BEVDetOcc · Ideal conditions Occ3D | mIoU: 44.5 -> 50.5 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- ProtoOcc — ProtoOcc: Accurate, Efficient 3D Occupancy Prediction Using Dual Branch Encoder-Prototype Query Decoder (AAAI 2025) · better on 1 of 1 shared result
- COTR — COTR: Correspondence Transformer for Matching Across Images (ICCV 2021) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Achieving Speed-Accuracy Balance in Vision-based 3D Occupancy Prediction via Geometric-Semantic Disentanglement (AAAI 2025 · similarity 0.89)
- 3D Occupancy Prediction with Low-Resolution Queries via Prototype-aware View Transformation (CVPR 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 0.88)
- CSV-Occ: Fusing Multi-frame Alignment for Occupancy Prediction with Temporal Cross State Space Model and Central Voting Mechanism (ICML 2025 · similarity 0.88)
- Semi-supervised 3D Semantic Scene Completion with 2D Vision Foundation Model Guidance (AAAI 2025 · 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.
