Research map / Vision-language and multimodal / Human and embodied modeling
Hyperbolic Hierarchical Alignment Reasoning Network for Text-3D Retrieval
Wenrui Li, Yidan Lu, Yeyu Chai, Rui Zhao, Hengyu Man, Xiaopeng Fan
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 hierarchical ordering loss (ablation) | H2ARN · T3DR-HIT v2 | Rsum: 229.6 -> 238.5 | Improved |
| add contribution-aware hyperbolic aggregation (ablation) | H2ARN · T3DR-HIT v2 | Rsum: 233.5 -> 238.5 | Improved |
| add both ordering loss and aggregation (ablation) | H2ARN · T3DR-HIT v2 | Rsum: 222.0 -> 238.5 | Improved |
| use contribution-aware aggregation instead of mean pooling in Euclidean space (ablation) | Euclidean baseline · T3DR-HIT v2 | Rsum: 196.3 -> 215.1 | Improved |
| move from Euclidean to Lorentz hyperbolic space (with ordering loss) (ablation) | Eu + CA · T3DR-HIT v2 | Rsum: 215.1 -> 238.5 | Improved |
| replace CLIP+PointNet with CLIP+DGCNN backbone in RMARN | RMARN · T3DR-HIT v2 | Rsum: 127.3 -> 220.3 | Improved |
| vary attention heads and SA layers (ablation) | H2ARN · T3DR-HIT v2 | Rsum: 210.2 to 238.5 | Mixed |
| H2ARN vs RMARN with same CLIP+DGCNN backbones | H2ARN · T3DR-HIT v2 | Rsum: 220.3 -> 238.5 | Improved |
Closest papers by meaning
- Riemann-based Multi-scale Attention Reasoning Network for Text-3D Retrieval (AAAI 2025 · similarity 0.89)
- HyperSDFusion: Bridging Hierarchical Structures in Language and Geometry for Enhanced 3D Text2Shape Generation (CVPR 2024 · similarity 0.87)
- Retrieval-Augmented Score Distillation for Text-to-3D Generation (ICML 2024 · similarity 0.87)
- Escaping Plato's Cave: Towards the Alignment of 3D and Text Latent Spaces (CVPR 2025 · similarity 0.87)
- Tango3D: Towards Alignment for Global and Local 2D-3D Correspondence (NeurIPS 2026 · similarity 0.86)
- Accept the Modality Gap: An Exploration in the Hyperbolic Space (CVPR 2024 · similarity 0.85)
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.
