Research map / 3D vision / 3D generation
VEON: Vocabulary-Enhanced Occupancy Prediction
Jilai Zheng, Pin Tang, Zhongdao Wang, Guoqing Wang, 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 |
|---|---|---|---|
| train depth module with LoRA for domain transfer (ablation) | VEON-B · Occ3D-nuScenes val | mIoU: 11.57 -> 12.38 | Improved |
| train depth module with LoRA for domain transfer (ablation) | VEON-L · Occ3D-nuScenes val | mIoU: 14.57 -> 15.14 | Improved |
| add class reweighting loss (ablation) | VEON-B · Occ3D-nuScenes val | mIoU: 10.39 -> 12.38 | Improved |
| add class reweighting loss (ablation) | VEON-L · Occ3D-nuScenes val | mIoU: 14.18 -> 15.14 | Improved |
| add high-resolution side adaptor to CLIP (ablation) | VEON-B · Occ3D-nuScenes val | mIoU: 11.68 -> 12.38 | Improved |
| add attention bias matrix A in HSA (ablation) | VEON-B · Occ3D-nuScenes val | mIoU: 11.75 -> 12.38 | Improved |
| add supplementary matrix S in HSA (ablation) | VEON-B · Occ3D-nuScenes val | mIoU: 11.61 -> 12.38 | Improved |
| use attention bias instead of linear token offsets for CLIP tokens (ablation) | VEON-B · Occ3D-nuScenes val | mIoU: 12.06 -> 12.38 | Improved |
| add manual labels for 13 seen classes (ablation) | VEON-L · Occ3D-nuScenes val | mIoU: 15.14 -> 19.94 | Improved |
| add manual labels for 9 seen classes (ablation) | VEON-L · Occ3D-nuScenes val | mIoU: 15.14 -> 15.16 | No clear effect |
| use VEON-L instead of POP-3D for open-vocabulary retrieval | retrieval benchmark · Open-vocab retrieval test (all) | mAP: 12.6 -> 30.9 | Improved |
Closest papers by meaning
- LOMA: Language-assisted Semantic Occupancy Network via Triplane Mamba (AAAI 2025 · similarity 0.90)
- OccMamba: Semantic Occupancy Prediction with State Space Models (CVPR 2025 · similarity 0.89)
- AdaOcc: Adaptive 3D Occupancy Prediction for Embodied Tasks (NeurIPS 2026 · similarity 0.88)
- OctOcc: High-Resolution 3D Occupancy Prediction with Octree (AAAI 2024 · similarity 0.88)
- Achieving Speed-Accuracy Balance in Vision-based 3D Occupancy Prediction via Geometric-Semantic Disentanglement (AAAI 2025 · similarity 0.88)
- RadOcc: Learning Cross-Modality Occupancy Knowledge through Rendering Assisted Distillation (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.
