Korea AeroSpace Administration to Release 2025 Satellite Data Big Data AI Training Dataset

[Financial News] The Korea AeroSpace Administration (KASA) announced on the 21st that it will release the '2025 Satellite Data Big Data AI Training Dataset,' which utilizes national satellite data, comprising approximately 150,000 cases.
This dataset has been designed and developed with a focus on ocean monitoring and land cover change detection, both of which have recently drawn significant attention in the satellite industry and AI research. The Korea Aerospace Research Institute (KARI), the lead research institution for KASA’s ‘Satellite Data Big Data Utilization Support System Development Project,’ was responsible for constructing the dataset.
Since 2023, KASA has systematically built and released over 630,000 satellite data AI training datasets through the ‘Satellite Data Big Data Utilization Support System Development Project.’ With this new release, the range of applications available to industry and researchers will be further expanded. The large-scale AI training dataset, based on imagery from Korea Multi-Purpose Satellite-3 (KOMPSAT-3), Korea Multi-Purpose Satellite-3A, and Korea Multi-Purpose Satellite-5 (KOMPSAT-5), is expected to enhance the accuracy of satellite image analysis and promote the development of innovative technologies.
The newly released dataset consists of approximately 150,000 cases, including six ocean monitoring datasets (aquaculture farms, oil spills, coastlines, vessels, landfast ice, and sea ice) and a land cover time-series change dataset.
In addition to this new dataset, KASA has developed and released large-scale datasets for object detection (880,000 images), building segmentation (350,000 images), roads (over 10,000 km), cloud detection (7,500 images), and land cover classification (1,200 images), which are being utilized across various fields. All released datasets are conveniently accessible through the National Research Data Platform (DataON) operated by the Korea Institute of Science and Technology Information (KISTI).
Changheon Han, Director General of the Aerospace Industry Bureau at KASA, stated, “AI training datasets utilizing national satellite imagery will significantly enhance the use of satellite images by domestic researchers and industry. Furthermore, they will contribute to strengthening the global competitiveness of artificial intelligence and space technology. We will continue to proactively address new demands by consistently building and releasing AI training datasets based on satellite imagery.”


jiany@fnnews.com Yeon Ji-an Reporter