Research map / Vision-language and multimodal / Human and embodied modeling
ORD: Object-Relation Decoupling for Generalized 3D Visual Grounding
Ronggang Huang, Fansen Meng, Huaidong Zhang, Xuemiao Xu
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 anchors and relative object relations (full vs w/o) (ablation) | ORD · Nr3D | Overall accuracy (%): 62.2 -> 71.6 | Improved |
| add SFAM global and predicate alignment modules (ablation) | ORD · Nr3D | Overall accuracy (%): 69.1 -> 71.6 | Improved |
| add predicate-decoupled alignment (PAM) (ablation) | ORD · Nr3D | Overall accuracy (%): 70.8 -> 71.6 | Improved |
| add SIEM and AIEM fusion modules (ablation) | ORD · Nr3D | Overall accuracy (%): 70.9 -> 71.6 | Improved |
| add anchor-guided regression module (ablation) | ORD · Nr3D | Overall accuracy (%): 71.0 -> 71.6 | Improved |
| add localization loss L_ref (ablation) | ORD · Nr3D | Overall accuracy (%): 62.2 -> 64.3 | Improved |
| add relation loss L_r on top of L_ref (ablation) | ORD · Nr3D | Overall accuracy (%): 64.3 -> 70.4 | Improved |
| add feature-to-relation and text-to-relation contrastive losses (ablation) | ORD · Nr3D | Overall accuracy (%): 70.4 -> 71.1 (f2r), 71.3 (t2r) | Improved |
| ORD vs ViewSRD on relation 'above' | ORD · Nr3D (viewpoint-independent) | accuracy (%): 80.4 -> 70.5 | Hurt |
| ORD vs ViewSRD on view-dependent split | ORD · Sr3D | View-dependent accuracy (%): 69.0 -> 64.2 | Hurt |
| ORD vs ViewSRD overall | ORD · Nr3D | Overall accuracy (%): 69.9 -> 71.6 | Improved |
Closest papers by meaning
- AugRefer: Advancing 3D Visual Grounding via Cross-Modal Augmentation and Spatial Relation-based Referring (AAAI 2025 · similarity 0.92)
- Towards CLIP-driven Language-free 3D Visual Grounding via 2D-3D Relational Enhancement and Consistency (CVPR 2024 · similarity 0.92)
- G^3-LQ: Marrying Hyperbolic Alignment with Explicit Semantic-Geometric Modeling for 3D Visual Grounding (CVPR 2024 · similarity 0.91)
- From Objects to Anywhere: A Holistic Benchmark for Multi-level Visual Grounding in 3D Scenes (NeurIPS 2025 · similarity 0.90)
- Omni-Q: Omni-Directional Scene Understanding for Unsupervised Visual Grounding (CVPR 2024 · similarity 0.90)
- Connecting the Dots: Training-Free Visual Grounding via Agentic Reasoning (AAAI 2026 · similarity 0.89)
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.
