Research map / 3D vision / 3D generation
Monocular Open Vocabulary Occupancy Prediction for Indoor Scenes
Changqing Zhou, Yueru Luo, Han Zhang, Zeyu Jiang, Changhao Chen
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 GaussianFormer2 aggregation with Poisson G2O (ablation) | LegoOcc · Occ-ScanNet (closed-vocab) | IoU: 56.96 -> 59.24 | Improved |
| replace GaussianFormer2 aggregation with Poisson G2O (ablation) | LegoOcc · Occ-ScanNet (open-vocab) | mIoU: 0.00 -> 21.05 | Improved |
| replace Bernoulli aggregation with Poisson G2O (ablation) | LegoOcc · Occ-ScanNet (open-vocab) | IoU: +12.85 IoU, +3.80 mIoU | Improved |
| replace Bernoulli aggregation with Poisson G2O (ablation) | LegoOcc · Occ-ScanNet (closed-vocab) | mIoU: 51.68 -> 52.80 | Improved |
| add temperature schedule (exponential decay) versus fixed temperature 1 (ablation) | LegoOcc · Occ-ScanNet (open-vocab) | mIoU: 18.15 -> 21.05 | Improved |
| use exponential instead of linear temperature decay (ablation) | LegoOcc · Occ-ScanNet (open-vocab) | mIoU: 2.30 -> 21.05 | Improved |
| match test temperature to training end temperature rather than 1 vs 1e-3 mismatch (ablation) | LegoOcc · Occ-ScanNet (open-vocab) | IoU: 32.52 -> 59.50 | Improved |
| lower Tmin from 1e-3 to 1e-4 (ablation) | LegoOcc · Occ-ScanNet (open-vocab) | mIoU: 21.05 -> 20.37 | Hurt |
| raise Tmin from 1e-3 to 1e-2 (ablation) | LegoOcc · Occ-ScanNet (open-vocab) | mIoU: 21.05 -> 20.85 | Hurt |
| use geometry-only open-vocab LegoOcc instead of re-implemented LOcc | monocular open-vocab occupancy · Occ-ScanNet (open-vocab) | mIoU: 9.25 -> 21.05 | Improved |
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.
- EmbodiedOcc — EmbodiedOcc: Embodied 3D Occupancy Prediction for Vision-based Online Scene Understanding (ICCV 2025) · 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
- OccMamba: Semantic Occupancy Prediction with State Space Models (CVPR 2025 · similarity 0.90)
- OctOcc: High-Resolution 3D Occupancy Prediction with Octree (AAAI 2024 · similarity 0.90)
- Achieving Speed-Accuracy Balance in Vision-based 3D Occupancy Prediction via Geometric-Semantic Disentanglement (AAAI 2025 · similarity 0.89)
- LOMA: Language-assisted Semantic Occupancy Network via Triplane Mamba (AAAI 2025 · similarity 0.89)
- OctreeOcc: Efficient and Multi-Granularity Occupancy Prediction Using Octree Queries (NeurIPS 2024 · similarity 0.89)
- GaussianFormer-2: Probabilistic Gaussian Superposition for Efficient 3D Occupancy Prediction (CVPR 2025 · similarity 0.89)
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.
