2026-10-12T13:52:39Z https://www.scidb.cn/oai

10.57760/sciencedb.radars.00081 2026-08-17T11:20:20Z

2026-08-17 AIRSAT1.0: Space-Air Collaborative Multimodal Remote Sensing Interpretation Dataset doi:10.57760/sciencedb.radars.00081 en Intelligent interpretation of remote sensing imagery is an essential technological foundation for applications such as land cover monitoring, agricultural resource surveys, ecological environment assessment, and urban-rural development management. To meet the requirements of complex multi-object and multi-source collaborative interpretation in high-standard agricultural demonstration areas, this study constructs the Aerospace-Airborne Collaborative Multimodal Remote Sensing Interpretation Benchmark Dataset for Complex Objects in High-Standard Agricultural Demonstration Areas (AIRSAT1.0). The dataset is built upon observation data acquired by the AIRSAT satellite constellation independently developed by China Satellite, which consists of four satellites currently in orbit (three SAR satellites and one optical satellite). It integrates three data modalities: high-resolution SAR satellite imagery, optical satellite imagery, and unmanned aerial vehicle (UAV) remote sensing imagery. The data were collected from the High-Standard Agricultural Demonstration Area in Laiwu District, Jinan City, Shandong Province, China.The dataset contains eight typical land-cover categories, including wheat, corn, greenhouses, roads, power transmission facilities, water conservancy facilities, water bodies, and buildings, with a total of 170,610 image samples. It is characterized by aerospace observation collaboration, broad land-cover diversity, multi-scale sample distribution, and clear spatial correspondence between optical and SAR imagery.Benchmark experiments conducted on the optical-SAR registration subset demonstrate that all baseline methods achieve better semantic segmentation performance when using dual-modal inputs compared with single-modal inputs, with the highest mIoU improvement reaching 4.67%. Furthermore, UAV imagery experiments indicate that the dataset can effectively support fine-grained segmentation of high-resolution land objects and recognition of small-scale infrastructure targets.AIRSAT1.0 provides a valuable data foundation and evaluation benchmark for research on intelligent interpretation of aerospace collaborative multimodal remote sensing imagery. Multimodal remote sensing dataset; Optical-SAR-UAV multimodal fusion; Semantic segmentation; Object detection chenjie PUBLIC https://creativecommons.org/licenses/by-nc-nd/4.0/ dataset Science Data Bank