Papers › Deep Evidential Learning with Noisy Correspondence for Cross-Modal Retrieval

Deep Evidential Learning with Noisy Correspondence for Cross-Modal Retrieval

10 Oct 2022ACM International Conference on Multimedia 2022 10archive 2025-07-28

Yang Qin, Dezhong Peng, Xi Peng, Xu Wang, Peng Hu

Cross-modal retrieval has been a compelling topic in the multimodal community. Recently, to mitigate the high cost of data collection, the co-occurred pairs (e.g., image and text) could be collected from the Internet as a large-scaled cross-modal dataset, e.g., Conceptual Captions. However, it will unavoidably introduce noise (i.e., mismatched pairs) into training data, dubbed noisy correspondence. Unquestionably, such noise will make supervision information unreliable/uncertain and remarkably degrade the performance. Besides, most existing methods focus training on hard negatives, which will amplify the unreliability of noise. To address the issues, we propose a generalized Deep Evidential Cross-modal Learning framework (DECL), which integrates a novel Cross-modal Evidential Learning paradigm (CEL) and a Robust Dynamic Hinge loss (RDH) with positive and negative learning. CEL could capture and learn the uncertainty brought by noise to improve the robustness and reliability of cross-modal retrieval. Specifically, the bidirectional evidence based on cross-modal similarity is first modeled and parameterized into the Dirichlet distribution, which not only provides accurate uncertainty estimation but also imparts resilience to perturbations against noisy correspondence. To address the amplification problem, RDH smoothly increases the hardness of negatives focused on, thus embracing higher robustness against high noise. Extensive experiments are conducted on three image-text benchmark datasets, i.e., Flickr30K, MS-COCO, and Conceptual Captions, to verify the effectiveness and efficiency of the proposed method. The code is available at \urlhttps://github.com/QinYang79/DECL.

PaperPDFCode

Code

qinyang79/decl mentioned in paperpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

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

Tasks

Cross-Modal RetrievalCross-modal retrieval with noisy correspondenceRetrievalText-based Person Retrieval with Noisy Correspondence

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Cross-modal retrieval with noisy correspondence CC152K DECL-SGRAF Image-to-text R@1 39.0 #13 of 15 Archive leaderboard report
Cross-modal retrieval with noisy correspondence CC152K DECL-SGRAF Image-to-text R@10 75.5 #13 of 15 Archive leaderboard report
Cross-modal retrieval with noisy correspondence CC152K DECL-SGRAF Image-to-text R@5 66.1 #13 of 15 Archive leaderboard report
Cross-modal retrieval with noisy correspondence CC152K DECL-SGRAF R-Sum 364.3 #13 of 15 Archive leaderboard report
Cross-modal retrieval with noisy correspondence CC152K DECL-SGRAF Text-to-image R@1 40.7 #13 of 15 Archive leaderboard report
Cross-modal retrieval with noisy correspondence CC152K DECL-SGRAF Text-to-image R@10 76.7 #13 of 15 Archive leaderboard report
Cross-modal retrieval with noisy correspondence CC152K DECL-SGRAF Text-to-image R@5 66.3 #13 of 15 Archive leaderboard report
Cross-modal retrieval with noisy correspondence COCO-Noisy DECL-SGARF Image-to-text R@1 77.5 #16 of 17 Archive leaderboard report
Cross-modal retrieval with noisy correspondence COCO-Noisy DECL-SGARF Image-to-text R@10 98.4 #16 of 17 Archive leaderboard report
Cross-modal retrieval with noisy correspondence COCO-Noisy DECL-SGARF Image-to-text R@5 95.9 #16 of 17 Archive leaderboard report
Cross-modal retrieval with noisy correspondence COCO-Noisy DECL-SGARF R-Sum 518.2 #16 of 17 Archive leaderboard report
Cross-modal retrieval with noisy correspondence COCO-Noisy DECL-SGARF Text-to-image R@1 61.7 #16 of 17 Archive leaderboard report
Cross-modal retrieval with noisy correspondence COCO-Noisy DECL-SGARF Text-to-image R@10 95.4 #16 of 17 Archive leaderboard report
Cross-modal retrieval with noisy correspondence COCO-Noisy DECL-SGARF Text-to-image R@5 89.3 #16 of 17 Archive leaderboard report
Cross-modal retrieval with noisy correspondence Flickr30K-Noisy DECL-SGRAF Image-to-text R@1 77.5 #15 of 16 Archive leaderboard report
Cross-modal retrieval with noisy correspondence Flickr30K-Noisy DECL-SGRAF Image-to-text R@10 97.0 #15 of 16 Archive leaderboard report
Cross-modal retrieval with noisy correspondence Flickr30K-Noisy DECL-SGRAF Image-to-text R@5 93.8 #15 of 16 Archive leaderboard report
Cross-modal retrieval with noisy correspondence Flickr30K-Noisy DECL-SGRAF R-Sum 494.7 #15 of 16 Archive leaderboard report
Cross-modal retrieval with noisy correspondence Flickr30K-Noisy DECL-SGRAF Text-to-image R@1 56.1 #15 of 16 Archive leaderboard report
Cross-modal retrieval with noisy correspondence Flickr30K-Noisy DECL-SGRAF Text-to-image R@10 88.5 #15 of 16 Archive leaderboard report
Cross-modal retrieval with noisy correspondence Flickr30K-Noisy DECL-SGRAF Text-to-image R@5 81.8 #15 of 16 Archive leaderboard report
Text-based Person Retrieval with Noisy Correspondence CUHK-PEDES DECL Rank 10 91.93 #2 of 6 Archive leaderboard report
Text-based Person Retrieval with Noisy Correspondence CUHK-PEDES DECL Rank-1 70.29 #2 of 6 Archive leaderboard report
Text-based Person Retrieval with Noisy Correspondence CUHK-PEDES DECL Rank-5 87.04 #2 of 6 Archive leaderboard report
Text-based Person Retrieval with Noisy Correspondence CUHK-PEDES DECL mAP 62.84 #2 of 6 Archive leaderboard report
Text-based Person Retrieval with Noisy Correspondence CUHK-PEDES DECL mINP 46.54 #2 of 6 Archive leaderboard report
Text-based Person Retrieval with Noisy Correspondence ICFG-PEDES DECL Rank 1 61.95 #2 of 6 Archive leaderboard report
Text-based Person Retrieval with Noisy Correspondence ICFG-PEDES DECL Rank-10 83.88 #2 of 6 Archive leaderboard report
Text-based Person Retrieval with Noisy Correspondence ICFG-PEDES DECL Rank-5 78.36 #2 of 6 Archive leaderboard report
Text-based Person Retrieval with Noisy Correspondence ICFG-PEDES DECL mAP 36.08 #2 of 6 Archive leaderboard report
Text-based Person Retrieval with Noisy Correspondence ICFG-PEDES DECL mINP 6.25 #2 of 6 Archive leaderboard report
Text-based Person Retrieval with Noisy Correspondence RSTPReid DECL Rank 1 61.75 #2 of 6 Archive leaderboard report
Text-based Person Retrieval with Noisy Correspondence RSTPReid DECL Rank 10 86.90 #2 of 6 Archive leaderboard report
Text-based Person Retrieval with Noisy Correspondence RSTPReid DECL Rank 5 80.70 #2 of 6 Archive leaderboard report
Text-based Person Retrieval with Noisy Correspondence RSTPReid DECL mAP 47.70 #2 of 6 Archive leaderboard report
Text-based Person Retrieval with Noisy Correspondence RSTPReid DECL mINP 26.07 #2 of 6 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