CaltechAUTHORS
  A Caltech Library Service

Benchmarking Representation Learning for Natural World Image Collections

Van Horn, Grant and Cole, Elijah and Beery, Sara and Wilber, Kimberly and Belongie, Serge and MacAodha, Oisin (2021) Benchmarking Representation Learning for Natural World Image Collections. In: 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). IEEE , Piscataway, NJ, pp. 12879-12888. ISBN 978-1-6654-4509-2. https://resolver.caltech.edu/CaltechAUTHORS:20220105-847441800

[img] PDF - Submitted Version
See Usage Policy.

2MB

Use this Persistent URL to link to this item: https://resolver.caltech.edu/CaltechAUTHORS:20220105-847441800

Abstract

Recent progress in self-supervised learning has resulted in models that are capable of extracting rich representations from image collections without requiring any explicit label supervision. However, to date the vast majority of these approaches have restricted themselves to training on standard benchmark datasets such as ImageNet. We argue that fine-grained visual categorization problems, such as plant and animal species classification, provide an informative testbed for self-supervised learning. In order to facilitate progress in this area we present two new natural world visual classification datasets, iNat2021 and NeWT. The former consists of 2.7M images from 10k different species up-loaded by users of the citizen science application iNaturalist. We designed the latter, NeWT, in collaboration with domain experts with the aim of benchmarking the performance of representation learning algorithms on a suite of challenging natural world binary classification tasks that go beyond standard species classification. These two new datasets allow us to explore questions related to large-scale representation and transfer learning in the context of fine-grained categories. We provide a comprehensive analysis of feature extractors trained with and without supervision on ImageNet and iNat2021, shedding light on the strengths and weaknesses of different learned features across a diverse set of tasks. We find that features produced by standard supervised methods still outperform those produced by self-supervised approaches such as SimCLR. However, improved self-supervised learning methods are constantly being released and the iNat2021 and NeWT datasets are a valuable resource for tracking their progress.


Item Type:Book Section
Related URLs:
URLURL TypeDescription
https://doi.org/10.1109/cvpr46437.2021.01269DOIArticle
https://arxiv.org/abs/2103.16483arXivDiscussion Paper
ORCID:
AuthorORCID
Beery, Sara0000-0002-2544-1844
Additional Information:© 2021 IEEE. Thanks to the iNaturalist team and community for providing access to data, Eliot Miller and Mitch Barry for helping to curate NeWT, and to Pietro Perona for valuable feedback.
DOI:10.1109/cvpr46437.2021.01269
Record Number:CaltechAUTHORS:20220105-847441800
Persistent URL:https://resolver.caltech.edu/CaltechAUTHORS:20220105-847441800
Official Citation:G. Van Horn, E. Cole, S. Beery, K. Wilber, S. Belongie and O. MacAodha, "Benchmarking Representation Learning for Natural World Image Collections," 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021, pp. 12879-12888, doi: 10.1109/CVPR46437.2021.01269
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:112727
Collection:CaltechAUTHORS
Deposited By: Tony Diaz
Deposited On:09 Jan 2022 21:29
Last Modified:09 Jan 2022 21:29

Repository Staff Only: item control page