Research map / Vision-language and multimodal / Human and embodied modeling
Multi-branch Collaborative Learning Network for 3D Visual Grounding
Zhipeng Qian, Yiwei Ma, Zhekai Lin, Jiayi Ji, Xiawu Zheng, Xiaoshuai Sun, Rongrong Ji
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 |
|---|---|---|---|
| align with query mask instead of bounding box (ablation) | MCLN · ScanRefer | 3DREC Acc@0.5: 44.23 -> 45.53 | Improved |
| fuse relative position by addition instead of multiplication (ablation) | RSA module · ScanRefer | 3DREC Acc@0.5: 44.22 -> 45.53 | Improved |
| add Adaptive Soft Alignment (ASA) to base two-branch model (ablation) | MCLN · ScanRefer | 3DREC Acc@0.5: 42.39 -> 45.29 | Improved |
| add Relative Superpoint Aggregation (RSA) on top of ASA (ablation) | MCLN · ScanRefer | 3DRES mIoU: 42.85 -> 44.72 | Improved |
| add point-wise weights to Focal loss (ablation) | ASA module · ScanRefer | 3DREC Acc@0.5: 42.39 -> 45.02 | Improved |
| add mask-wise weights to Dice loss (ablation) | ASA module · ScanRefer | 3DREC Acc@0.5: 42.39 -> 44.76 | Improved |
| use adaptive weighted Focal/Dice instead of standard losses (ablation) | ASA module · ScanRefer | 3DRES mIoU: 44.24 -> 44.72 | Improved |
| add ASA module to reduce prediction conflicts between branches (ablation) | MCLN · ScanRefer | 3DIE (inconsistency): 4.97% -> 2.33% | Improved |
| use MCLN in single-stage setting on unique scenes | MCLN · ScanRefer unique, single-stage | Acc@0.25: 86.40 (EDA) -> 84.43 | Hurt |
| MCLN vs 3DRefTR on SR3D | MCLN · SR3D | Acc@0.25: 68.5 -> 68.4 | No clear effect |
Closest papers by meaning
- AugRefer: Advancing 3D Visual Grounding via Cross-Modal Augmentation and Spatial Relation-based Referring (AAAI 2025 · similarity 0.88)
- CoT3DRef: Chain-of-Thoughts Data-Efficient 3D Visual Grounding (ICLR 2024 · similarity 0.87)
- Multi-task Visual Grounding with Coarse-to-Fine Consistency Constraints (AAAI 2025 · similarity 0.87)
- LESS: Label-Efficient and Single-Stage Referring 3D Segmentation (NeurIPS 2024 · similarity 0.86)
- 3D-STMN: Dependency-Driven Superpoint-Text Matching Network for End-to-End 3D Referring Expression Segmentation (AAAI 2024 · similarity 0.86)
- X-RefSeg3D: Enhancing Referring 3D Instance Segmentation via Structured Cross-Modal Graph Neural Networks (AAAI 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.
