Reproduce “Dual-Modal Attention-Enhanced Text-Video Retrieval with Triplet Partial Margin Contrastive Learning” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Chen Jiang, Hong Liu, Xuzheng Yu, Qing Wang, Yuan Cheng, Jia Xu, Zhongyi Liu, Qingpei Guo and 3 more
The authors introduce a text–video retrieval method that identifies hard negatives using textual and visual attention and trains on masked versions of matched pairs, encouraging complete pairs to score higher than partial ones. On MSR-VTT with CLIP ViT-B/16, the full method raises text-to-video rsum from 207.8 to 212.2; R@1 is not highest, and results are benchmark-bound.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.