Research map / 3D vision / 3D generation
OccFormer: Dual-path Transformer for Vision-based 3D Semantic Occupancy Prediction
Yunpeng Zhang, Zheng Zhu, Dalong Du
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 the global path to the local path in the dual-path encoder (ablation) | OccFormer encoder · SemanticKITTI val | SSC mIoU: 12.95 -> 13.46 | Improved |
| replace 3D ResNet-16 with dual-path transformer encoder (ablation) | OccFormer · SemanticKITTI val | SSC mIoU: 12.89 -> 13.46 | Improved |
| use multi-scale 3D deformable attention pixel decoder instead of FPN-3D (ablation) | OccFormer · SemanticKITTI val | SSC mIoU: 12.89 -> 13.46 | Improved |
| increase pixel decoder layers from 3 to 6 (ablation) | OccFormer pixel decoder · SemanticKITTI val | SSC mIoU: 13.22 -> 13.46 | Improved |
| use max-pooling (preserve-pooling) instead of trilinear for attention mask downsampling (ablation) | OccFormer decoder · SemanticKITTI val | SSC mIoU: 11.61 -> 12.13 | Improved |
| use class-guided sampling instead of uniform sampling (ablation) | OccFormer decoder · SemanticKITTI val | SSC mIoU: 11.61 -> 13.01 | Improved |
| combine max-pool and class-guided sampling (ablation) | OccFormer decoder · SemanticKITTI val | SSC mIoU: 13.01 -> 13.46 | Improved |
| OccFormer vs MonoScene on SC IoU for validation | OccFormer · SemanticKITTI val | SC IoU: 36.86 -> 36.50 | No clear effect |
| OccFormer vs TPVFormer for LiDAR segmentation from camera | OccFormer · nuScenes test | mIoU: 69.4 -> 70.8 | Improved |
Later papers that compared with this method
18 later accepted papers report OccFormer as a baseline in their results tables; 15 of them report a better number on most of the results they share with it.
- OneOcc: Semantic Occupancy Prediction for Legged Robots with a Single Panoramic Camera (CVPR 2026 · better on 3 of 3 shared results)
- DLWM: Dual Latent World Models enable Holistic Gaussian-centric Pre-training in Autonomous Driving (CVPR 2026 · better on 1 of 1 shared result)
- Sparsity-Aware Voxel Attention and Foreground Modulation for 3D Semantic Scene Completion (CVPR 2026 · better on 1 of 1 shared result)
- GaussianFormer-2: Probabilistic Gaussian Superposition for Efficient 3D Occupancy Prediction (CVPR 2025 · better on 1 of 2 shared results)
- Monocular Semantic Scene Completion via Masked Recurrent Networks (ICCV 2025 · better on 2 of 2 shared results)
- ProtoOcc: Accurate, Efficient 3D Occupancy Prediction Using Dual Branch Encoder-Prototype Query Decoder (AAAI 2025 · better on 1 of 1 shared result)
- Spatiotemporal Decoupling for Efficient Vision-Based Occupancy Forecasting (CVPR 2025 · better on 1 of 1 shared result)
- SA-Occ: Satellite-Assisted 3D Occupancy Prediction in Real World (ICCV 2025 · better on 1 of 1 shared result)
10 more not listed.
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- MonoScene — MonoScene: Monocular 3D Semantic Scene Completion (CVPR 2022) · better on 2 of 2 shared results
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
- COTR: Compact Occupancy TRansformer for Vision-based 3D Occupancy Prediction (CVPR 2024 · similarity 0.92)
- Achieving Speed-Accuracy Balance in Vision-based 3D Occupancy Prediction via Geometric-Semantic Disentanglement (AAAI 2025 · similarity 0.91)
- SDGOCC: Semantic and Depth-Guided Bird's-Eye View Transformation for 3D Multimodal Occupancy Prediction (CVPR 2025 · similarity 0.91)
- CSV-Occ: Fusing Multi-frame Alignment for Occupancy Prediction with Temporal Cross State Space Model and Central Voting Mechanism (ICML 2025 · similarity 0.90)
- OctOcc: High-Resolution 3D Occupancy Prediction with Octree (AAAI 2024 · similarity 0.89)
- ODG: Occupancy Prediction Using Dual Gaussians (NeurIPS 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.
