SimMAT: Exploring Transferability from Vision Foundation Models to Any Image Modality

Chenyang Lei, Liyi Chen, Jun Cen, Xiao Chen, Zhen Lei, Felix Heide, Ziwei Liu, Qifeng Chen, Zhaoxiang Zhang
{"title":"SimMAT: Exploring Transferability from Vision Foundation Models to Any Image Modality","authors":"Chenyang Lei, Liyi Chen, Jun Cen, Xiao Chen, Zhen Lei, Felix Heide, Ziwei Liu, Qifeng Chen, Zhaoxiang Zhang","doi":"arxiv-2409.08083","DOIUrl":null,"url":null,"abstract":"Foundation models like ChatGPT and Sora that are trained on a huge scale of\ndata have made a revolutionary social impact. However, it is extremely\nchallenging for sensors in many different fields to collect similar scales of\nnatural images to train strong foundation models. To this end, this work\npresents a simple and effective framework SimMAT to study an open problem: the\ntransferability from vision foundation models trained on natural RGB images to\nother image modalities of different physical properties (e.g., polarization).\nSimMAT consists of a modality-agnostic transfer layer (MAT) and a pretrained\nfoundation model. We apply SimMAT to a representative vision foundation model\nSegment Anything Model (SAM) to support any evaluated new image modality. Given\nthe absence of relevant benchmarks, we construct a new benchmark to evaluate\nthe transfer learning performance. Our experiments confirm the intriguing\npotential of transferring vision foundation models in enhancing other sensors'\nperformance. Specifically, SimMAT can improve the segmentation performance\n(mIoU) from 22.15% to 53.88% on average for evaluated modalities and\nconsistently outperforms other baselines. We hope that SimMAT can raise\nawareness of cross-modal transfer learning and benefit various fields for\nbetter results with vision foundation models.","PeriodicalId":501130,"journal":{"name":"arXiv - CS - Computer Vision and Pattern Recognition","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2024-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - CS - Computer Vision and Pattern Recognition","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2409.08083","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0

Abstract

Foundation models like ChatGPT and Sora that are trained on a huge scale of data have made a revolutionary social impact. However, it is extremely challenging for sensors in many different fields to collect similar scales of natural images to train strong foundation models. To this end, this work presents a simple and effective framework SimMAT to study an open problem: the transferability from vision foundation models trained on natural RGB images to other image modalities of different physical properties (e.g., polarization). SimMAT consists of a modality-agnostic transfer layer (MAT) and a pretrained foundation model. We apply SimMAT to a representative vision foundation model Segment Anything Model (SAM) to support any evaluated new image modality. Given the absence of relevant benchmarks, we construct a new benchmark to evaluate the transfer learning performance. Our experiments confirm the intriguing potential of transferring vision foundation models in enhancing other sensors' performance. Specifically, SimMAT can improve the segmentation performance (mIoU) from 22.15% to 53.88% on average for evaluated modalities and consistently outperforms other baselines. We hope that SimMAT can raise awareness of cross-modal transfer learning and benefit various fields for better results with vision foundation models.
SimMAT:探索从视觉基础模型到任何图像模式的可移植性
像 ChatGPT 和 Sora 这样的基础模型是在大量数据的基础上训练出来的,已经产生了革命性的社会影响。然而,对于许多不同领域的传感器来说,要收集类似规模的自然图像来训练强大的基础模型是极具挑战性的。为此,本研究提出了一个简单有效的框架 SimMAT 来研究一个开放性问题:从在自然 RGB 图像上训练的视觉基础模型到不同物理特性(如偏振)的其他图像模态的可转移性。我们将 SimMAT 应用于具有代表性的视觉基础模型--分段任意模型(SAM),以支持任何经过评估的新图像模态。鉴于缺乏相关基准,我们构建了一个新的基准来评估迁移学习性能。我们的实验证实了转移视觉基础模型在提高其他传感器性能方面的巨大潜力。具体来说,SimMAT 可以将所评估模态的分割性能(mIoU)从平均 22.15% 提高到 53.88%,并持续优于其他基线。我们希望 SimMAT 能够提高人们对跨模态迁移学习的认识,并使各个领域受益,从而利用视觉基础模型获得更好的结果。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信