Saved in:
Bibliographic Details
Main Authors: Cha, Keumgang, Yu, Donggeun, Seo, Junghoon
Format: Preprint
Published: 2024
Subjects:
Online Access:https://arxiv.org/abs/2409.07048
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909311366594560
author Cha, Keumgang
Yu, Donggeun
Seo, Junghoon
author_facet Cha, Keumgang
Yu, Donggeun
Seo, Junghoon
contents The prominence of generalized foundation models in vision-language integration has witnessed a surge, given their multifarious applications. Within the natural domain, the procurement of vision-language datasets to construct these foundation models is facilitated by their abundant availability and the ease of web crawling. Conversely, in the remote sensing domain, although vision-language datasets exist, their volume is suboptimal for constructing robust foundation models. This study introduces an approach to curate vision-language datasets by employing an image decoding machine learning model, negating the need for human-annotated labels. Utilizing this methodology, we amassed approximately 9.6 million vision-language paired datasets in VHR imagery. The resultant model outperformed counterparts that did not leverage publicly available vision-language datasets, particularly in downstream tasks such as zero-shot classification, semantic localization, and image-text retrieval. Moreover, in tasks exclusively employing vision encoders, such as linear probing and k-NN classification, our model demonstrated superior efficacy compared to those relying on domain-specific vision-language datasets.
format Preprint
id arxiv_https___arxiv_org_abs_2409_07048
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Pushing the Limits of Vision-Language Models in Remote Sensing without Human Annotations
Cha, Keumgang
Yu, Donggeun
Seo, Junghoon
Computer Vision and Pattern Recognition
The prominence of generalized foundation models in vision-language integration has witnessed a surge, given their multifarious applications. Within the natural domain, the procurement of vision-language datasets to construct these foundation models is facilitated by their abundant availability and the ease of web crawling. Conversely, in the remote sensing domain, although vision-language datasets exist, their volume is suboptimal for constructing robust foundation models. This study introduces an approach to curate vision-language datasets by employing an image decoding machine learning model, negating the need for human-annotated labels. Utilizing this methodology, we amassed approximately 9.6 million vision-language paired datasets in VHR imagery. The resultant model outperformed counterparts that did not leverage publicly available vision-language datasets, particularly in downstream tasks such as zero-shot classification, semantic localization, and image-text retrieval. Moreover, in tasks exclusively employing vision encoders, such as linear probing and k-NN classification, our model demonstrated superior efficacy compared to those relying on domain-specific vision-language datasets.
title Pushing the Limits of Vision-Language Models in Remote Sensing without Human Annotations
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2409.07048