Research map / 3D vision / 3D generation
FLM-Occ: Feed-forward Likelihood Maximization for Efficient Indoor Occupancy Prediction
Guangcheng Chen, Lihuang Fang, Huaqi Tao, Yicheng He, Li He, Zhang Hong
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 |
|---|---|---|---|
| fine-tune the DAv2 ViT (relative features) (ablation) | FLM-Occ (128 superquadrics) · Occ-ScanNet | mIoU: 43.9 -> 59.2 | Improved |
| fine-tune the DAv2 ViT (metric features) (ablation) | FLM-Occ (128 superquadrics) · Occ-ScanNet | mIoU: 48.9 -> 59.6 | Improved |
| use metric DAv2 instead of relative (frozen ViT) (ablation) | FLM-Occ (128 superquadrics) · Occ-ScanNet | mIoU: 43.9 -> 48.9 | Improved |
| use metric DAv2 instead of relative (finetuned ViT) (ablation) | FLM-Occ (128 superquadrics) · Occ-ScanNet | mIoU: 59.2 -> 59.6 | No clear effect |
| add density regularization loss L_reg (ablation) | FLM-Occ (128 superquadrics) · Occ-ScanNet | mIoU: 59.6 -> 62.0 | Improved |
| increase refinement blocks from 2 to 4 (ablation) | FLM-Occ (128 superquadrics) · Occ-ScanNet | mIoU: 60.7 -> 62.0 | Improved |
| increase refinement blocks from 4 to 5 (ablation) | FLM-Occ (128 superquadrics) · Occ-ScanNet | mIoU: 62.0 -> 61.8 | No clear effect |
| scale encoder from DAv2-ViT-B to ViT-L (ablation) | FLM-Occ (128 superquadrics) · Occ-ScanNet | mIoU: 62.0 -> 65.0 | Improved |
| use superquadrics instead of Gaussians (32 primitives) (ablation) | FLM-Occ · Occ-ScanNet | mIoU: 48.8 -> 55.9 | Improved |
| use superquadrics instead of Gaussians (1024 primitives) (ablation) | FLM-Occ · Occ-ScanNet | mIoU: 63.6 -> 64.0 | Improved |
| increase primitives from 32 to 1024 (superquadrics) (ablation) | FLM-Occ · Occ-ScanNet | mIoU: 55.9 -> 64.0 | Improved |
| increase primitives from 32 to 1024 (superquadrics) (ablation) | FLM-Occ · Occ-ScanNet | FPS: 32.1 -> 25.4 | Hurt |
Closest papers by meaning
- OPUS: Occupancy Prediction Using a Sparse Set (NeurIPS 2024 · similarity 0.89)
- OccMamba: Semantic Occupancy Prediction with State Space Models (CVPR 2025 · similarity 0.87)
- ODG: Occupancy Prediction Using Dual Gaussians (NeurIPS 2025 · similarity 0.86)
- LowRankOcc: Tensor Decomposition and Low-Rank Recovery for Vision-based 3D Semantic Occupancy Prediction (CVPR 2024 · similarity 0.86)
- Spatiotemporal Decoupling for Efficient Vision-Based Occupancy Forecasting (CVPR 2025 · similarity 0.86)
- OctreeOcc: Efficient and Multi-Granularity Occupancy Prediction Using Octree Queries (NeurIPS 2024 · similarity 0.86)
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.
