Research map / Vision-language and multimodal / Human and embodied modeling
POP-3D: Open-Vocabulary 3D Occupancy Prediction from Images
Antonín Vobecký, Oriane Siméoni, David Hurych, Spyros Gidaris, Andrei Bursuc, Patrick Perez, Josef Sivic
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 detailed nuScenes class descriptions as text queries instead of class names (ablation) | POP-3D · nuScenes val 25%, 16 classes | mIoU: 10.2 -> 23.0 | Improved |
| replace original-5 queries with manmade-5 queries (ablation) | POP-3D · nuScenes val 25%, 5 classes | mIoU: 27.5 -> 34.7 | Improved |
| replace original-5 queries with pedestrian-5 queries (ablation) | POP-3D · nuScenes val 25%, 5 classes | mIoU: 27.5 -> 35.0 | Improved |
| distill into 3D instead of directly transferring MaskCLIP+ features | POP-3D · nuScenes LiDAR points | LiDAR mIoU: 23.0 -> 26.4 | Improved |
| POP-3D vs MaskCLIP+ for text-query 3D retrieval | POP-3D · Retrieval benchmark | mAP: 14.9 -> 18.4 | Improved |
| class-agnostic occupancy head instead of supervised TPVFormer training | POP-3D · nuScenes val | class-agnostic IoU: +11.5 points | Improved |
| zero-shot POP-3D vs fully supervised TPVFormer | POP-3D · nuScenes val | (16+1)-class mIoU: approx 78% | Hurt |
| lower loss weight lambda from 1.00 to 0.25 (ablation) | POP-3D · nuScenes half-res RN50 | mIoU: 12.0 -> 11.9 | No clear effect |
| lower loss weight lambda from 1.00 to 0.50 (ablation) | POP-3D · nuScenes half-res RN50 | IoU: 30.0 -> 30.5 | No clear effect |
| RN50 to RN101 backbone (ablation) | POP-3D · nuScenes half-res | mIoU: 12.0 -> 15.1 | Improved |
| half to full input resolution with RN50 (ablation) | POP-3D · nuScenes | mIoU: 12.0 -> 12.3 | No clear effect |
| half to full input resolution with RN101 (ablation) | POP-3D · nuScenes | mIoU: 15.1 -> 15.2 | No clear effect |
Later papers that compared with this method
1 later accepted paper reports POP-3D as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- AGO: Adaptive Grounding for Open World 3D Occupancy Prediction (ICCV 2025 · better on 5 of 5 shared results)
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.
- MaskCLIP+ — MaskCLIP: Masked Self-Distillation Advances Contrastive Language-Image Pretraining (CVPR 2023) · better on 2 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
- LOMA: Language-assisted Semantic Occupancy Network via Triplane Mamba (AAAI 2025 · similarity 0.89)
- Open-Vocabulary 3D Semantic Segmentation with Foundation Models (CVPR 2024 · similarity 0.89)
- LowRankOcc: Tensor Decomposition and Low-Rank Recovery for Vision-based 3D Semantic Occupancy Prediction (CVPR 2024 · similarity 0.89)
- Segment Any 3D Object with Language (ICLR 2025 · similarity 0.89)
- OpenLex3D: A Tiered Benchmark for Open-Vocabulary 3D Scene Representations (NeurIPS 2025 · similarity 0.89)
- Mosaic3D: Foundation Dataset and Model for Open-Vocabulary 3D Segmentation (CVPR 2025 · 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.
