Breaking Through the Noisy Correspondence: A Robust Model for Image-Text Matching

Author:

Shi Haitao1ORCID,Liu Meng2ORCID,Mu Xiaoxuan1ORCID,Song Xuemeng3ORCID,Hu Yupeng1ORCID,Nie Liqiang4ORCID

Affiliation:

1. Shandong University, Jinan, China

2. Shandong Jianzhu University, Jinan, China

3. Shandong University, Qingdao, China

4. Harbin Institute of Technology (Shenzhen), Shenzhen, China

Abstract

Unleashing the power of image-text matching in real-world applications is hampered by noisy correspondence. Manually curating high-quality datasets is expensive and time-consuming, and datasets generated using diffusion models are not adequately well-aligned. The most promising way is to collect image-text pairs from the Internet, but it will inevitably introduce noisy correspondence. To reduce the negative impact of noisy correspondence, we propose a novel model that first transforms the noisy correspondence filtering problem into a similarity distribution modeling problem by exploiting the powerful capabilities of pre-trained models. Specifically, we use the Gaussian Mixture model to model the similarity obtained by CLIP as clean distribution and noisy distribution, to filter out most of the noisy correspondence in the dataset. Afterward, we used relatively clean data to fine-tune the model. To further reduce the negative impact of unfiltered noisy correspondence, i.e., a minimal part where two distributions intersect during the fine-tuning process, we propose a distribution-sensitive dynamic margin ranking loss, further increasing the distance between the two distributions. Through continuous iteration, the noisy correspondence gradually decreases and the model performance gradually improves. Our extensive experiments demonstrate the effectiveness and robustness of our model even under high noise rates.

Funder

National Natural Science Foundation of China

Shandong Provincial Natural Science Foundation

Science and Technology Innovation Program for Distinguished Young Scholars of Shandong Province Higher Education Institutions

Publisher

Association for Computing Machinery (ACM)

Reference61 articles.

1. Eric Arazo, Diego Ortego, Paul Albert, Noel E. O’Connor, and Kevin McGuinness. 2019. Unsupervised Label Noise Modeling and Loss Correction. In Proceedings of the International Conference on Machine Learning, Vol. 97. 312–321.

2. Global Relation-Aware Attention Network for Image-Text Retrieval

3. IMRAM: Iterative Matching With Recurrent Attention Memory for Cross-Modal Image-Text Retrieval

4. Cross-modal Graph Matching Network for Image-text Retrieval

5. Junyoung Chung Caglar Gülcehre KyungHyun Cho and Yoshua Bengio. 2014. Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling. arXiv:2201.08239. Retrieved from https://arxiv.org/abs/1412.3555

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3