Krishna, Tarun, McGuinness, Kevin ORCID: 0000-0003-1336-6477 and O'Connor, Noel E. ORCID: 0000-0002-4033-9135 (2021) Evaluating contrastive models for instance-based image retrieval. In: ACM International Conference on Multimedia Retrieval, 21-24 Aug 2021, Taipei, Taiwan. ISBN 978-1-4503-8463-6
Abstract
In this work, we evaluate contrastive models for the task of imageretrieval. We hypothesise that models that are learned to encodesemantic similarity among instances via discriminative learningshould perform well on the task of image retrieval, where rele-vancy is defined in terms of instances of the same object. Throughour extensive evaluation, we find that representations from mod-els trained using contrastive methods perform on-par with (andoutperforms) a pre-trained supervised baseline trained on the Ima-geNet labels in retrieval tasks under various configurations. This isremarkable given that the contrastive models require no explicitsupervision. Thus, we conclude that these models can be used tobootstrap base models to build more robust image retrieval engines.
Metadata
Item Type: | Conference or Workshop Item (Paper) |
---|---|
Event Type: | Conference |
Refereed: | Yes |
Uncontrolled Keywords: | Contrastive Learning; Instance Retrieval; Deep Learning |
Subjects: | UNSPECIFIED |
DCU Faculties and Centres: | DCU Faculties and Schools > Faculty of Engineering and Computing > School of Electronic Engineering Research Institutes and Centres > INSIGHT Centre for Data Analytics |
Published in: | ICMR '21: Proceedings of the 2021 International Conference on Multimedia Retrieval. . Association for Computing Machinery (ACM). ISBN 978-1-4503-8463-6 |
Publisher: | Association for Computing Machinery (ACM) |
Official URL: | https://doi.org/10.1145/3460426.3463585 |
Copyright Information: | © 2021 The Authors (CC-BY-4.0) |
ID Code: | 25806 |
Deposited On: | 07 Sep 2021 13:53 by Tarun Krishna . Last Modified 09 Nov 2021 16:10 |
Documents
Full text available as:
Preview |
PDF
- Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader
538kB |
Downloads
Downloads
Downloads per month over past year
Archive Staff Only: edit this record