Research map / 3D vision / 3D generation
OccGen: Generative Multi-modal 3D Occupancy Prediction for Autonomous Driving
Guoqing Wang, Zhongdao Wang, Pin Tang, Jilai Zheng, Xiangxuan Ren, Bailan Feng, Chao Ma
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 |
|---|---|---|---|
| replace discriminative one-shot occupancy prediction with diffusion-based noise-to-occupancy decoder (multi-modal) | CONet · nuScenes-Occupancy val (multi-modal) | mIoU: 20.1 -> 22.0 | Improved |
| apply generative OccGen on camera-only input | C-CONet · nuScenes-Occupancy val (camera-only) | mIoU: 12.8 -> 14.5 | Improved |
| apply OccGen to camera-based SemanticKITTI | OccFormer · SemanticKITTI val | mIoU: 13.46 -> 13.74 | Improved |
| OccGen IoU on SemanticKITTI vs VoxFormer | OccGen · SemanticKITTI val | IoU: 44.02 -> 36.87 | Hurt |
| add the progressive refinement (generative) decoder (ablation) | baseline occupancy network · nuScenes-Occupancy multi-modal | mIoU: 20.4 -> 21.6 | Improved |
| add the proposed multi-modal conditional encoder only (ablation) | baseline occupancy network · nuScenes-Occupancy multi-modal | mIoU: 20.4 -> 20.7 | Improved |
| add 3D deformable cross-attention and self-attention (ablation) | OccGen decoder · nuScenes-Occupancy multi-modal | mIoU: 20.7 -> 22.0 | Improved |
| add diffusion time module (ablation) | OccGen decoder · nuScenes-Occupancy multi-modal | mIoU: 21.7 -> 22.0 | Improved |
| add hard 2D-to-3D view transformation alone (ablation) | OccGen encoder · nuScenes-Occupancy multi-modal | mIoU: 21.4 -> 21.5 | No clear effect |
| add both hard LSS and geometry mask (ablation) | OccGen encoder · nuScenes-Occupancy multi-modal | mIoU: 21.4 -> 22.0 | Improved |
| increase sampling steps from 1 to 2 (multi-modal) (ablation) | OccGen · nuScenes-Occupancy multi-modal | mIoU: 21.7 -> 21.8 | Improved |
| OccGen latency vs CONet at one sampling step (multi-modal) | OccGen · nuScenes-Occupancy multi-modal | latency (ms): 344.8 -> 357.1 | Hurt |
Later papers that compared with this method
3 later accepted papers report OccGen as a baseline in their results tables; 3 of them report a better number on most of the results they share with it.
- LightOcc: Lightweight Spatial Embedding for Efficient Vision-based 3D Occupancy Prediction (IJCV 2026 · better on 2 of 2 shared results)
- SPSC: Sparse and Scalable Multi-Modal 3D Occupancy Prediction for Autonomous Driving (AAAI 2026 · better on 1 of 1 shared result)
- OmniNWM: Unifying the State-Action-Reward Triad for Closed-Loop Panoramic Driving Navigation World Models (ECCV 2026 · better on 1 of 1 shared result)
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.
- OccFormer — OccFormer: Dual-path Transformer for Vision-based 3D Semantic Occupancy Prediction (ICCV 2023) · 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 0.92)
- OctOcc: High-Resolution 3D Occupancy Prediction with Octree (AAAI 2024 · similarity 0.92)
- OPUS: Occupancy Prediction Using a Sparse Set (NeurIPS 2024 · 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 0.91)
- ODG: Occupancy Prediction Using Dual Gaussians (NeurIPS 2025 · similarity 0.90)
- UniScene: Unified Occupancy-centric Driving Scene Generation (CVPR 2025 · similarity 0.90)
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.
