Research map / 3D vision / 3D generation
Streaming Dense Voxel Representations for 3D Occupancy Prediction
Seokha Moon, Janghyun Baek, Yujin Jeong, Daewon Chae, Giseop Kim, Jungbeom Lee, Jinkyu Kim, Sunwook Choi
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 |
|---|---|---|---|
| add StreamAgg temporal voxel aggregation to the single-frame base (ablation) | StreamOcc base · Occ3D-nuScenes | mIoU: 36.8 -> 40.4 | Improved |
| add an auxiliary detection head to StreamAgg (ablation) | StreamAgg · Occ3D-nuScenes | mIoU_D: 35.4 -> 36.3 | Improved |
| add QueryAgg to StreamAgg (ablation) | StreamAgg · Occ3D-nuScenes | mIoU_D: 35.4 -> 38.1 | Improved |
| add Adaptive Residual Refinement to naive voxel streaming (ablation) | Naive voxel streaming · Occ3D-nuScenes | mIoU: 38.72 -> 39.84 | Improved |
| add Semantic Supervision to the refinement module (ablation) | StreamAgg refinement · Occ3D-nuScenes | mIoU: 39.84 -> 40.25 | Improved |
| add Geometry Supervision to the refinement module (ablation) | StreamAgg refinement · Occ3D-nuScenes | mIoU: 40.25 -> 40.37 | Improved |
| add voxel-to-query detector (V2Q) alone to StreamAgg (ablation) | StreamAgg · Occ3D-nuScenes | mIoU: 40.37 -> 40.06 | Hurt |
| add image-to-query detector (I2Q) alone to StreamAgg (ablation) | StreamAgg · Occ3D-nuScenes | mIoU: 40.37 -> 40.42 | No clear effect |
| add V2Q on top of I2Q (ablation) | StreamAgg + I2Q · Occ3D-nuScenes | mIoU: 40.42 -> 40.78 | Improved |
| add Dynamic Query Aggregation (DQA) on top of I2Q+V2Q (ablation) | StreamAgg + I2Q + V2Q · Occ3D-nuScenes | mIoU: 40.78 -> 41.90 | Improved |
| replace QueryAgg with Spatial Cross-Attention for image-to-voxel aggregation (ablation) | StreamAgg · Occ3D-nuScenes | mIoU_D: 38.12 -> 35.77 | Hurt |
| use QueryAgg instead of Spatial Cross-Attention on latency (ablation) | StreamAgg · Occ3D-nuScenes | latency (ms): 95.2 -> 83.3 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 4 on most shared results.
- GaussianWorld — GaussianWorld: Gaussian World Model for Streaming 3D Occupancy Prediction (CVPR 2025) · better on 2 of 2 shared results
- STCOcc — STCOcc: Sparse Spatial-Temporal Cascade Renovation for 3D Occupancy and Scene Flow Prediction (CVPR 2025) · better on 0 of 2 shared results
- ViewFormer — ViewFormer: NeRF-Free Neural Rendering from Few Images Using Transformers (ECCV 2022) · better on 2 of 2 shared results
- COTR — COTR: Correspondence Transformer for Matching Across Images (ICCV 2021) · better on 0 of 2 shared results
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
- 3D Occupancy Prediction with Low-Resolution Queries via Prototype-aware View Transformation (CVPR 2025 · similarity 0.89)
- ProtoOcc: Accurate, Efficient 3D Occupancy Prediction Using Dual Branch Encoder-Prototype Query Decoder (AAAI 2025 · similarity 0.89)
- VoxelSplat: Dynamic Gaussian Splatting as an Effective Loss for Occupancy and Flow Prediction (CVPR 2025 · similarity 0.87)
- OctOcc: High-Resolution 3D Occupancy Prediction with Octree (AAAI 2024 · similarity 0.86)
- ODG: Occupancy Prediction Using Dual Gaussians (NeurIPS 2025 · similarity 0.86)
- ViPOcc: Leveraging Visual Priors from Vision Foundation Models for Single-View 3D Occupancy Prediction (AAAI 2025 · 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.
