CR-CLIP通过动态交互提升跨模态检索精度，显著优于基线。
ContextRefine-CLIP for EPIC-KITCHENS-100 Multi-Instance Retrieval Challenge 2025
- 引入交叉模态注意力流，实现视觉与文本特征双向动态优化。
- 在EPIC-KITCHENS-100上达66.78mAP、82.08nDCG，无需集成学习。
- 适合关注跨模态对齐与高效检索的研究者或应用开发者。
本文提出ContextRefine-CLIP（CR-CLIP），一种用于视觉-文本多实例检索的高效模型。基于双编码器AVION架构，引入交叉模态注意力流模块，实现视觉与文本特征间的双向动态交互与精炼，生成更具上下文感知能力的联合表征。针对如EPIC-KITCHENS-100中提供的软标签相关性矩阵，CR-CLIP结合对称多相似性损失，利用精炼特征实现更精确的语义对齐与优化。不使用集成学习，在EPIC-KITCHENS-100公开排行榜上取得66.78mAP和82.08nDCG的成绩，显著超越基线模型，充分验证了其在跨模态检索中的有效性。代码将开源发布于https://github.com/delCayr/ContextRefine-Clip。
原文摘要 · Abstract (English)
This report presents ContextRefine-CLIP (CR-CLIP), an efficient model for visual-textual multi-instance retrieval tasks. The approach is based on the dual-encoder AVION, on which we introduce a cross-modal attention flow module to achieve bidirectional dynamic interaction and refinement between visual and textual features to generate more context-aware joint representations. For soft-label relevance matrices provided in tasks such as EPIC-KITCHENS-100, CR-CLIP can work with Symmetric Multi-Similarity Loss to achieve more accurate semantic alignment and optimization using the refined features. Without using ensemble learning, the CR-CLIP model achieves 66.78mAP and 82.08nDCG on the EPIC-KITCHENS-100 public leaderboard, which significantly outperforms the baseline model and fully validates its effectiveness in cross-modal retrieval. The code will be released open-source on https://github.com/delCayr/ContextRefine-Clip
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。