Research map / Point clouds / Segmentation and detection
QueryOcc: Query-based Self-Supervision for 3D Semantic Occupancy
Adam Lilja, Ji Lan, Junsheng Fu, Lars Hammarstrand
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 |
|---|---|---|---|
| use query-based 4D supervision instead of rendering-based supervision (ablation) | QueryOcc · Occ3D-nuScenes | Semantic RayIoU: 15.0 -> 23.6 | Improved |
| add VFM feature distillation (ablation) | QueryOcc · Occ3D-nuScenes | Semantic RayIoU: 23.6 -> 24.0 | Improved |
| add rendering losses to query-based supervision (ablation) | QueryOcc · Occ3D-nuScenes | Semantic RayIoU: 23.6 -> 23.3 | No clear effect |
| add log-linear bins, long supervision, contraction and point encoding to LSS (ablation) | QueryOcc · Occ3D-nuScenes | Semantic RayIoU: 21.7 -> 23.6 | Improved |
| use lidar supervision instead of pseudo point clouds (ablation) | QueryOcc · Occ3D-nuScenes | Semantic RayIoU: 23.6 -> 22.9 | Hurt |
| use lidar supervision instead of pseudo point clouds (ablation) | QueryOcc · Occ3D-nuScenes | Occupancy RayIoU: 45.2 -> 48.3 | Improved |
| combine pseudo and lidar supervision (ablation) | QueryOcc · Occ3D-nuScenes | Semantic RayIoU: 23.6 -> 24.6 | Improved |
| add Argoverse 2 training data (ablation) | QueryOcc (lidar) · Occ3D-nuScenes | Semantic RayIoU: 22.9 -> 26.0 | Improved |
| add Argoverse 2 training data (ablation) | QueryOcc (lidar) · Occ3D-nuScenes | Dynamic RayIoU: 19.4 -> 23.8 | Improved |
| subsample pseudo point cloud to fewer than 100k points (ablation) | QueryOcc · Occ3D-nuScenes | mRayIoU | No clear effect |
| raise input resolution from 256x256 to 900x1600 (ablation) | QueryOcc · Occ3D-nuScenes | mRayIoU: +3 mRayIoU (approx.) | Improved |
Closest papers by meaning
- SelfOcc: Self-Supervised Vision-Based 3D Occupancy Prediction (CVPR 2024 · similarity 0.91)
- Semi-supervised 3D Semantic Scene Completion with 2D Vision Foundation Model Guidance (AAAI 2025 · similarity 0.89)
- Sonata: Self-Supervised Learning of Reliable Point Representations (CVPR 2025 · similarity 0.88)
- AdaOcc: Adaptive 3D Occupancy Prediction for Embodied Tasks (NeurIPS 2026 · similarity 0.87)
- UnO: Unsupervised Occupancy Fields for Perception and Forecasting (CVPR 2024 · similarity 0.87)
- OctOcc: High-Resolution 3D Occupancy Prediction with Octree (AAAI 2024 · similarity 0.87)
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.
