Research map / Vision-language and multimodal / Cross-modal learning
Beyond Categorical Matching: Intra-Class Graded Relevance Estimation for Cross-Modal 3D Retrieval
shunning liu, YingPeng Zhang, Jianing Lin, Yifan Wang, Yang Zhang, Chun Yuan
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 SCI module (ablation) | ReMU3D · INGRE | NDCG@5: 0.57 -> 0.94 | Improved |
| add SCI module (ablation) | ReMU3D · INGRE | NDCG@10: 0.61 -> 0.93 | Improved |
| add Q-MoE module (ablation) | ReMU3D · INGRE | NDCG@5: 0.66 -> 0.94 | Improved |
| add Q-MoE module (ablation) | ReMU3D · INGRE | Kendall's tau: 0.24 -> 0.41 | Improved |
| add SCI module (ablation) | ReMU3D · INGRE | Spearman's rho: 0.22 -> 0.54 | Improved |
| joint training instead of two-stage training (ablation) | ReMU3D · INGRE | NDCG@5: 0.41 -> 0.94 | Improved |
| dynamic modality-ratio sampling instead of fixed ratio (ablation) | ReMU3D · INGRE | NDCG@5: 0.68 -> 0.94 | Improved |
| add MSE soft-label loss to InfoNCE (ablation) | ReMU3D · INGRE | NDCG@5: 0.94 -> 0.93 | No clear effect |
| add KL soft-label loss to InfoNCE (ablation) | ReMU3D · INGRE | NDCG@10: 0.93 -> 0.91 | Hurt |
| increase number of Q-MoE experts beyond moderate scale (ablation) | ReMU3D · INGRE | NDCG | No clear effect |
| use reparameterized multimodal encoder over Uni3D | ReMU3D · INGRE | NDCG@10: 0.73 -> 0.93 | Improved |
| use ReMU3D over Uni3D | ReMU3D · ScanObjectNN zero-shot | Top-1 accuracy: 66.5 -> 68.1 | Improved |
Closest papers by meaning
- Cross the Gap: Exposing the Intra-modal Misalignment in CLIP via Modality Inversion (ICLR 2025 · similarity 0.87)
- CrossOver: 3D Scene Cross-Modal Alignment (CVPR 2025 · similarity 0.86)
- DREAM: Decoupled Discriminative Learning with Bigraph-aware Alignment for Semi-supervised 2D-3D Cross-modal Retrieval (AAAI 2025 · similarity 0.86)
- Cross-Modal 3D Representation with Multi-View Images and Point Clouds (CVPR 2025 · similarity 0.86)
- Beyond Global Similarity: Multi-Conditional Retrieval for Fine-Grained Cross-Modal Understanding (CVPR 2026 · similarity 0.86)
- TAMM: TriAdapter Multi-Modal Learning for 3D Shape Understanding (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.
