Research map / Point clouds / Point cloud classification
PV-Ground: Text-Guided Point-Voxel Interaction for 3D Visual Grounding
Junpeng Shang, Feifei Shao, Jun Xiao, Lin Li, Hongwei Wang, Dongfang Ma
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 point-voxel interaction (voxel backbone with keypoint aggregation) (ablation) | MCLN-style point-based baseline · ScanRefer grounding (single-stage) | Acc@0.5: 42.64 -> 45.19 | Improved |
| add text-guided keypoint sampling in place of Top-K selection (ablation) | MCLN-style point-based baseline · ScanRefer grounding (single-stage) | Acc@0.5: 42.64 -> 47.23 | Improved |
| add TGS on top of PVI (ablation) | PV-Ground · ScanRefer grounding (single-stage) | Acc@0.5: 45.19 -> 47.77 | Improved |
| add PVI on top of TGS (ablation) | PV-Ground · ScanRefer grounding (single-stage) | Acc@0.5: 47.23 -> 47.77 | Improved |
| add both PVI and TGS (ablation) | MCLN-style point-based baseline · ScanRefer segmentation (single-stage) | mIoU: 43.49 -> 47.73 | Improved |
| replace MCLN point-based baseline with PV-Ground | single-stage pipeline · ScanRefer overall | Acc@0.25: 54.30 -> 59.31 | Improved |
| replace MCLN point-based baseline with PV-Ground | two-stage pipeline · ScanRefer overall | Acc@0.5: 45.53 -> 47.56 | Improved |
| replace MCLN with PV-Ground on Nr3D | single-stage pipeline · Nr3D | accuracy: 45.7 -> 51.3 | Improved |
| use PV-Ground instead of TSP3D on Sr3D single-stage | single-stage pipeline · Sr3D | accuracy: 57.1 -> 56.5 | Hurt |
| replace MCLN with PV-Ground for referring segmentation | ScanRefer segmentation · ScanRefer overall | mIoU: 44.7 -> 47.9 | Improved |
Closest papers by meaning
- SeeGround: See and Ground for Zero-Shot Open-Vocabulary 3D Visual Grounding (CVPR 2025 · similarity 0.89)
- Text-guided Sparse Voxel Pruning for Efficient 3D Visual Grounding (CVPR 2025 · similarity 0.89)
- From Objects to Anywhere: A Holistic Benchmark for Multi-level Visual Grounding in 3D Scenes (NeurIPS 2025 · similarity 0.88)
- SPAZER: Spatial-Semantic Progressive Reasoning Agent for Zero-shot 3D Visual Grounding (NeurIPS 2025 · similarity 0.87)
- Multi-Object 3D Grounding with Dynamic Modules and Language-Informed Spatial Attention (NeurIPS 2024 · similarity 0.87)
- AugRefer: Advancing 3D Visual Grounding via Cross-Modal Augmentation and Spatial Relation-based Referring (AAAI 2025 · 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.
