Causal Inference Hashing for Long-Tailed Image Retrieval.
Saved in:
| Title: | Causal Inference Hashing for Long-Tailed Image Retrieval. |
|---|---|
| Authors: | Jin, Lu1 lu.jin@njust.edu.cn, Lu, Zhengyun1 zhengyunlu@njust.edu.cn, Li, Zechao1 zechao.li@njust.edu.cn, Pan, Yonghua1 yonghuapan@njust.edu.cn, Dai, Longquan1 dailongquan@njust.edu.cn, Tang, Jinhui2 tangjh@njfu.edu.cn, Jain, Ramesh3 jain@ics.uci.edu |
| Source: | IEEE Transactions on Image Processing. 2025, Vol. 34, p5099-5114. 16p. |
| Subjects: | Image retrieval, Hashing, Information retrieval, Electronic file management, Multimedia systems |
| Abstract: | In hashing-based long-tailed image retrieval, the dominance of data-rich head classes often hinders the learning of effective hash codes for data-poor tail classes due to inherent long-tailed bias. Interestingly, this bias also contains valuable prior knowledge by revealing inter-class dependencies, which can be beneficial for hash learning. However, previous methods have not thoroughly analyzed this tangled negative and positive effects of long-tailed bias from a causal inference perspective. In this paper, we propose a novel hash framework that employs causal inference to disentangle detrimental bias effects from beneficial ones. To capture good bias in long-tailed datasets, we construct hash mediators that conserve valuable prior knowledge from class centers. Furthermore, we propose a de-biased hash loss To enhance the beneficial bias effects while mitigating adverse ones, leading to more discriminative hash codes. Specifically, this loss function leverages the beneficial bias captured by hash mediators to support accurate class label prediction, while mitigating harmful bias by blocking its causal path to the hash codes and refining predictions through backdoor adjustment. Extensive experimental results on four widely used datasets demonstrate that the proposed method improves retrieval performance against the state-of-the-art methods by large margins. The source code is available at https://github.com/IMAG-LuJin/CIH [ABSTRACT FROM AUTHOR] |
| Copyright of IEEE Transactions on Image Processing is the property of IEEE and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.) | |
| Database: | Engineering Source |
| FullText | Text: Availability: 0 |
|---|---|
| Header | DbId: egs DbLabel: Engineering Source An: 191897151 AccessLevel: 6 PubType: Academic Journal PubTypeId: academicJournal PreciseRelevancyScore: 0 |
| IllustrationInfo | |
| Items | – Name: Title Label: Title Group: Ti Data: Causal Inference Hashing for Long-Tailed Image Retrieval. – Name: Author Label: Authors Group: Au Data: <searchLink fieldCode="AR" term="%22Jin%2C+Lu%22">Jin, Lu</searchLink><relatesTo>1</relatesTo><i> lu.jin@njust.edu.cn</i><br /><searchLink fieldCode="AR" term="%22Lu%2C+Zhengyun%22">Lu, Zhengyun</searchLink><relatesTo>1</relatesTo><i> zhengyunlu@njust.edu.cn</i><br /><searchLink fieldCode="AR" term="%22Li%2C+Zechao%22">Li, Zechao</searchLink><relatesTo>1</relatesTo><i> zechao.li@njust.edu.cn</i><br /><searchLink fieldCode="AR" term="%22Pan%2C+Yonghua%22">Pan, Yonghua</searchLink><relatesTo>1</relatesTo><i> yonghuapan@njust.edu.cn</i><br /><searchLink fieldCode="AR" term="%22Dai%2C+Longquan%22">Dai, Longquan</searchLink><relatesTo>1</relatesTo><i> dailongquan@njust.edu.cn</i><br /><searchLink fieldCode="AR" term="%22Tang%2C+Jinhui%22">Tang, Jinhui</searchLink><relatesTo>2</relatesTo><i> tangjh@njfu.edu.cn</i><br /><searchLink fieldCode="AR" term="%22Jain%2C+Ramesh%22">Jain, Ramesh</searchLink><relatesTo>3</relatesTo><i> jain@ics.uci.edu</i> – Name: TitleSource Label: Source Group: Src Data: <searchLink fieldCode="JN" term="%22IEEE+Transactions+on+Image+Processing%22">IEEE Transactions on Image Processing</searchLink>. 2025, Vol. 34, p5099-5114. 16p. – Name: Subject Label: Subjects Group: Su Data: <searchLink fieldCode="DE" term="%22Image+retrieval%22">Image retrieval</searchLink><br /><searchLink fieldCode="DE" term="%22Hashing%22">Hashing</searchLink><br /><searchLink fieldCode="DE" term="%22Information+retrieval%22">Information retrieval</searchLink><br /><searchLink fieldCode="DE" term="%22Electronic+file+management%22">Electronic file management</searchLink><br /><searchLink fieldCode="DE" term="%22Multimedia+systems%22">Multimedia systems</searchLink> – Name: Abstract Label: Abstract Group: Ab Data: In hashing-based long-tailed image retrieval, the dominance of data-rich head classes often hinders the learning of effective hash codes for data-poor tail classes due to inherent long-tailed bias. Interestingly, this bias also contains valuable prior knowledge by revealing inter-class dependencies, which can be beneficial for hash learning. However, previous methods have not thoroughly analyzed this tangled negative and positive effects of long-tailed bias from a causal inference perspective. In this paper, we propose a novel hash framework that employs causal inference to disentangle detrimental bias effects from beneficial ones. To capture good bias in long-tailed datasets, we construct hash mediators that conserve valuable prior knowledge from class centers. Furthermore, we propose a de-biased hash loss To enhance the beneficial bias effects while mitigating adverse ones, leading to more discriminative hash codes. Specifically, this loss function leverages the beneficial bias captured by hash mediators to support accurate class label prediction, while mitigating harmful bias by blocking its causal path to the hash codes and refining predictions through backdoor adjustment. Extensive experimental results on four widely used datasets demonstrate that the proposed method improves retrieval performance against the state-of-the-art methods by large margins. The source code is available at https://github.com/IMAG-LuJin/CIH [ABSTRACT FROM AUTHOR] – Name: AbstractSuppliedCopyright Label: Group: Ab Data: <i>Copyright of IEEE Transactions on Image Processing is the property of IEEE and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract.</i> (Copyright applies to all Abstracts.) |
| PLink | https://search.ebscohost.com/login.aspx?direct=true&site=eds-live&db=egs&AN=191897151 |
| RecordInfo | BibRecord: BibEntity: Identifiers: – Type: doi Value: 10.1109/TIP.2025.3588054 Languages: – Code: eng Text: English PhysicalDescription: Pagination: PageCount: 16 StartPage: 5099 Subjects: – SubjectFull: Image retrieval Type: general – SubjectFull: Hashing Type: general – SubjectFull: Information retrieval Type: general – SubjectFull: Electronic file management Type: general – SubjectFull: Multimedia systems Type: general Titles: – TitleFull: Causal Inference Hashing for Long-Tailed Image Retrieval. Type: main BibRelationships: HasContributorRelationships: – PersonEntity: Name: NameFull: Jin, Lu – PersonEntity: Name: NameFull: Lu, Zhengyun – PersonEntity: Name: NameFull: Li, Zechao – PersonEntity: Name: NameFull: Pan, Yonghua – PersonEntity: Name: NameFull: Dai, Longquan – PersonEntity: Name: NameFull: Tang, Jinhui – PersonEntity: Name: NameFull: Jain, Ramesh IsPartOfRelationships: – BibEntity: Dates: – D: 01 M: 01 Text: 2025 Type: published Y: 2025 Identifiers: – Type: issn-print Value: 10577149 Numbering: – Type: volume Value: 34 Titles: – TitleFull: IEEE Transactions on Image Processing Type: main |
| ResultId | 1 |
