Research map / Vision-language and multimodal / Cross-modal learning
DSGCR: Decomposed Spectral Geometry-Aware Cross-Modal Semantic Representation for 3D Visual Grounding
Jing He, Licheng Jiao, Lingling Li, Xiaoqiang Lu, Xu Liu, Wenping Ma, Fang Liu, Long Sun
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 Text-Aware Feature Tuning (TFT) to frozen-encoder baseline (ablation) | DSGCR baseline · ScanRefer 3DREC/3DRES | Acc@0.5: 44.54 -> 45.74 | Improved |
| full fine-tune the visual backbone (ablation) | DSGCR baseline · ScanRefer 3DREC | Acc@0.5: 44.54 -> 44.99 | No clear effect |
| add DSG module alone (ablation) | DSGCR baseline · ScanRefer 3DRES | mIoU: 44.22 -> 45.35 | Improved |
| combine TFT and DSG (ablation) | DSGCR baseline · ScanRefer 3DREC | Acc@0.5: 44.54 -> 46.97 | Improved |
| add Dynamic Gated Adapter on top of TASC (ablation) | TFT · ScanRefer 3DREC | Acc@0.5: 44.76 -> 45.74 | Improved |
| add TASC on top of DGA (ablation) | TFT · ScanRefer 3DREC | Acc@0.5: 44.74 -> 45.74 | Improved |
| add symmetric component in DSG (ablation) | DSG · ScanRefer 3DREC | Acc@0.5: 44.54 -> 45.28 | Improved |
| add antisymmetric component to symmetric one in DSG (ablation) | DSG · ScanRefer 3DRES | mIoU: 44.68 -> 45.35 | Improved |
| replace linear DSG mapping with nonlinear MLP (ablation) | DSG · ScanRefer 3DREC | Acc@0.5: 45.70 -> 46.97 | Improved |
| reduce spectral latent dim d from 5 to 3 (ablation) | DSG · ScanRefer 3DREC | Acc@0.5: 46.97 -> 44.78 | Hurt |
| increase spectral latent dim d from 5 to 8 (ablation) | DSG · ScanRefer 3DREC/3DRES | Acc@0.5: 46.97 -> 45.98 (3DREC Acc@0.5); 45.68 -> 45.76 (mIoU) | Mixed |
| change frequency dim D from 32 to 16 or 64 (ablation) | DSG · ScanRefer 3DREC | Acc@0.5: 46.97 -> 44.96 (D=16); 46.97 -> 44.39 (D=64) | Hurt |
Closest papers by meaning
- G^3-LQ: Marrying Hyperbolic Alignment with Explicit Semantic-Geometric Modeling for 3D Visual Grounding (CVPR 2024 · similarity 0.89)
- CrossOver: 3D Scene Cross-Modal Alignment (CVPR 2025 · similarity 0.88)
- AugRefer: Advancing 3D Visual Grounding via Cross-Modal Augmentation and Spatial Relation-based Referring (AAAI 2025 · similarity 0.88)
- S$^2$-MLLM: Boosting Spatial Reasoning Capability of MLLMs for 3D Visual Grounding with Structural Guidance (CVPR 2026 · similarity 0.87)
- Towards CLIP-driven Language-free 3D Visual Grounding via 2D-3D Relational Enhancement and Consistency (CVPR 2024 · similarity 0.87)
- Cross-Modal and Uncertainty-Aware Agglomeration for Open-Vocabulary 3D Scene Understanding (CVPR 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.
