AEROSPACE SHANGHAI, Volume. 42, Issue 2, 9(2025)
Review on the Development of Aerospace Remote Sensing Foundation Models and Prospects for Their Industrial Application
Figures & Tables(9)
Fig. 1. The release trend of open-source remote sensing training data
Fig. 2. The growth trend of the parameter number for aerospace remote sensing foundation models
Fig. 3. An example of cross-modal interaction with aerospace remote sensing foundation models
Fig. 4. Typical tasks supported by aerospace remote sensing foundation models
Fig. 5. Application efficiency of aerospace remote sensing foundation models
Fig. 6. Farmland monitoring by the SenseEarth platform based on aerospace remote sensing foundation models
Fig. 7. Analysis on forest fire by the IBM watsonx.ai platform based on aerospace remote sensing foundation models
Fig. 8. Military decision support by the Palantir artificial intelligence platform based on aerospace remote sensing foundation models
Table 1.
A summary of research achievements with aerospace remote sensing foundation models View tableView in ArticleTable 1.
A summary of research achievements with aerospace remote sensing foundation models 类型 典型成果 介绍 航天遥感视觉大模型 RingMo[ 11 ]实现了百万级光学遥感图像的自监督学习，支持场景分类、目标检测、语义分割、变化检测等多任务应用 SpectralGPT[ 12 ]首个专为高光谱遥感数据设计的6亿参数大模型，实现了百万级高光谱遥感图像的自监督学习 SARATR-X [ 13 ]首个公开发表的SAR图像目标识别基础模型，提出了适用于SAR图像的自监督学习的新框架 SkySense[ 14 ]兼备可见光、红外、SAR等多模态时序遥感图像数据融合与跨任务通用解译能力，模型参数超过20亿 航天遥感视觉-语言大模型 RemoteCLIP[ 15 ]学习了具有丰富语义的鲁棒视觉特征，并与文本语言特征对齐，实现了遥感图文检索和目标计数等处理能力 LHRS-Bot[ 16 ]通过视觉-语言特征对齐策略，将遥感视觉知识融入大语言模型中，使其具备遥感图像视觉问答和定位等能力 RS-LLaVA[ 17 ]基于LLaVA大模型构建，通过LoRA（低秩适应）微调使其能够处理遥感图像语义描述和视觉问答任务 RingMoGPT[ 18 ]结合了视觉、语言和地理定位能力，能够处理遥感图像场景分类、目标检测、视觉问答、语义描述以及变化检测等任务 航天遥感图像生成大模型 DiffusionSat[ 19 ]将遥感图像元数据作为条件信息纳入扩散模型，能够根据元数据生成相应遥感图像，且支持超分辨率生成、图像修复等任务 MetaEarth[ 20 ]提出自级联生成框架和滑动窗口生成方法，实现了多种分辨率、无边界且覆盖全球任意地理位置的遥感图像生成 HSIGene[ 21 ]同时支持无条件、单条件和多条件可控生成与真实图像相当的高光谱遥感图像，支持高光谱遥感图像去噪和超分辨率生成任务 Text2Earth[ 22 ]可根据用户输入的文本描述和分辨率要求，生成无边界限制的全球范围内典型地理场景的遥感图像
Tools
Get Citation
Copy Citation Text
Boyi SHANGGUAN, Ying HE, Luyun TIAN, Pengming FENG, Mengke ZHU, Haiyi REN, Guangjun HE. Review on the Development of Aerospace Remote Sensing Foundation Models and Prospects for Their Industrial Application[J]. AEROSPACE SHANGHAI, 2025, 42(2): 9
Paper Information
Category: Research and Application of Large Models
Received: Feb. 24, 2025
Accepted: --
Published Online: May. 26, 2025
The Author Email:
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20