Papers › NAC: Mitigating Noisy Correspondence in Cross-Modal Matching Via Neighbor Auxiliary Corrector

NAC: Mitigating Noisy Correspondence in Cross-Modal Matching Via Neighbor Auxiliary Corrector

18 Mar 2024International Conference on Acoustics, Speech, and Signal Processing 2024 3archive 2025-07-28

Yuqing Li, Haoming Huang, Jian Xu, Shao-Lun Huang

The presence of noisy correspondence within cross-modal matching has significantly undermined the performance of existing matching methods. In this paper, we introduce a robust framework named Neighbor Auxiliary Corrector (NAC) for alleviating noise by utilizing the neighbors, which are indicative of similar textual targets. NAC is inspired by an observation that similar texts tend to correspond to similar images. Leveraging the zero-shot capabilities of Pre-trained Language Models (PLMs), we identify the top-k nearest neighbors for each positive image-text pair. Subsequently, the side information provided by these neighbors is harnessed for both sample verification and sample rectification. Extensive experiments on benchmark datasets demonstrate that our framework can significantly boost the performance and is more robust to various levels of noisy correspondence.

PaperPDF

Code

No code repository is listed for this paper in the archive or in Syntology's graph.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Cross-modal retrieval with noisy correspondence

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Cross-modal retrieval with noisy correspondence CC152K NAC Image-to-text R@1 41.8 #5 of 15 Archive leaderboard report
Cross-modal retrieval with noisy correspondence CC152K NAC Image-to-text R@10 77.3 #5 of 15 Archive leaderboard report
Cross-modal retrieval with noisy correspondence CC152K NAC Image-to-text R@5 68.6 #5 of 15 Archive leaderboard report
Cross-modal retrieval with noisy correspondence CC152K NAC R-Sum 373.5 #5 of 15 Archive leaderboard report
Cross-modal retrieval with noisy correspondence CC152K NAC Text-to-image R@1 40.5 #5 of 15 Archive leaderboard report
Cross-modal retrieval with noisy correspondence CC152K NAC Text-to-image R@10 77.0 #5 of 15 Archive leaderboard report
Cross-modal retrieval with noisy correspondence CC152K NAC Text-to-image R@5 68.3 #5 of 15 Archive leaderboard report
Cross-modal retrieval with noisy correspondence COCO-Noisy NAC Image-to-text R@1 80.3 #9 of 17 Archive leaderboard report
Cross-modal retrieval with noisy correspondence COCO-Noisy NAC Image-to-text R@10 98.5 #9 of 17 Archive leaderboard report
Cross-modal retrieval with noisy correspondence COCO-Noisy NAC Image-to-text R@5 96.2 #9 of 17 Archive leaderboard report
Cross-modal retrieval with noisy correspondence COCO-Noisy NAC R-Sum 524.5 #9 of 17 Archive leaderboard report
Cross-modal retrieval with noisy correspondence COCO-Noisy NAC Text-to-image R@1 63.2 #9 of 17 Archive leaderboard report
Cross-modal retrieval with noisy correspondence COCO-Noisy NAC Text-to-image R@10 96.0 #9 of 17 Archive leaderboard report
Cross-modal retrieval with noisy correspondence COCO-Noisy NAC Text-to-image R@5 90.3 #9 of 17 Archive leaderboard report
Cross-modal retrieval with noisy correspondence Flickr30K-Noisy NAC Image-to-text R@1 79.3 #5 of 16 Archive leaderboard report
Cross-modal retrieval with noisy correspondence Flickr30K-Noisy NAC Image-to-text R@10 97.8 #5 of 16 Archive leaderboard report
Cross-modal retrieval with noisy correspondence Flickr30K-Noisy NAC Image-to-text R@5 94.6 #5 of 16 Archive leaderboard report
Cross-modal retrieval with noisy correspondence Flickr30K-Noisy NAC R-Sum 507.1 #5 of 16 Archive leaderboard report
Cross-modal retrieval with noisy correspondence Flickr30K-Noisy NAC Text-to-image R@1 60.8 #5 of 16 Archive leaderboard report
Cross-modal retrieval with noisy correspondence Flickr30K-Noisy NAC Text-to-image R@10 90.1 #5 of 16 Archive leaderboard report
Cross-modal retrieval with noisy correspondence Flickr30K-Noisy NAC Text-to-image R@5 84.5 #5 of 16 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections