GEO DATA

Data Article

KOMPSAT-3/3A Image-text Dataset for Training Large Multimodal Models: Han Oh, Dong-Bin Shin, Dae-Won Chung; GEO DATA. 2025;7(1):27-35. Published online March 19, 2025; DOI: https://doi.org/10.22761/GD.2025.0003

322 View
25 Download

Abstract PDF: This study aims to improve the accuracy and interpretability of large multimodal models (LMMs) specialized in satellite image analysis by constructing an image-text dataset based on KOMPSAT-3/3A imagery and presenting the results of training using this dataset. Conventional LMMs are primarily trained on general images, limiting their ability to effectively interpret the specific characteristics of satellite imagery, such as spectral bands, spatial resolution, and viewing angles. To address this limitation, we developed an image-text dataset, divided into pretraining and finetuning stages, based on the existing KOMPSAT object detection dataset. The pretraining dataset consists of captions summarizing the overall theme and key information of each image. The fine-tuning dataset integrates metadata -including acquisition time, sensor type, and coordinates- with detailed object detection labels to generate six types of question-answer pairs: detailed descriptions, conversations with varying answer lengths, bounding box identification, multiple choice questions, and complex reasoning. This structured dataset enables the model to learn not only the general context of satellite images but also fine-grained details such as object quantity, location, and geographic attributes. Training with the new KOMPSAT-based dataset significantly improved the model’s accuracy in recognizing regional information and object characteristics in satellite imagery. Finetuned models achieved substantially higher accuracy than previous models, surpassing even the GPT-4o model and demonstrating the effectiveness of a domain-specific dataset. The findings of this study are expected to contribute to various remote sensing applications, including automated satellite image analysis, change detection, and object detection.

Original Paper

GeoAI Dataset for Rural Hazardous Facilities Segmentation from KOMPSAT Ortho Mosaic Imagery: Sung-Hyun Gong, Hyung-Sup Jung, Moung-Jin Lee, Kwang-Jae Lee, Kwan-Young Oh, Jae-Young Chang; GEO DATA. 2023;5(4):231-237. Published online December 28, 2023; DOI: https://doi.org/10.22761/GD.2023.0054

2,234 View
83 Download
1 Citations

In South Korea, rural areas have been recognized for their potential as sustainable spaces for the future, but they are currently facing major problems. Unplanned construction of facilities such as factories, livestock facilities, and solar panels near residential areas is destroying the rural environment and deteriorating the quality of life of residents. Detection and monitoring of rural facilities are necessary to prevent disorderly development in rural areas and to manage rural space in a planned manner. In this study, satellite imagery data was utilized to obtain information on rural areas, which is useful for observing large areas and monitoring time series changes compared to field surveys. In this study, KOMPSAT ortho-mosaic optical imagery from 2019 and 2020 were utilized to construct AI training datasets for rural hazardous facilities segmentation for Seosan, Anseong, Naju, and Geochang areas. The dataset can be used in image segmentation models to classify rural facilities and can be used to monitor potentially hazardous facilities in rural areas. It is expected to contribute to solving rural problems by serving as the basis for rural planning.

Citations

Citations to this article as recorded by

Performance Comparison of Water Body Detection from Sentinel-1 SAR and Sentinel-2 Optical Imagery Using Attention U-Net Model
Il-Hoon Choi, Eu-Ru Lee, Hyung-Sup Jung
Korean Journal of Remote Sensing.2024; 40(5-1): 507. CrossRef

Articles

GEO-KOMPSAT-2A/2B AMI/GOCI-II/GEMS Data & Products: Sungsik Huh, Kyoung-Wook Jin; GEO DATA. 2022;4(4):39-49. Published online December 31, 2022; DOI: https://doi.org/10.22761/DJ2022.4.4.005

2,558 View
107 Download
1 Citations

Abstract PDF

Two geostationary satellites developed by the Korea Aerospace Research Institute and currently in operation are the GEO-KOMPSAT-2A (GK-2A) and the GEO-KOMPSAT-2B (GK-2B). The main instruments mounted on these satellites are the Advanced Meteorological Imager (AMI), the Geostationary Ocean Color Imager (GOCI-II) and the Geostationary Environment Monitoring Spectrometer (GEMS). This paper briefly introduced the GK-2A and GK-2B programs including measurement principles and elements of the instruments. Moreover, the data formats, operational products, and applications are summarized.

Citations

Citations to this article as recorded by

Hyperspectral Image-Based Identification of Maritime Objects Using Convolutional Neural Networks and Classifier Models
Dongmin Seo, Daekyeom Lee, Sekil Park, Sangwoo Oh
Journal of Marine Science and Engineering.2024; 13(1): 6. CrossRef

KOMPSAT Optical Image Data Provision and Quality Management: Daesoon Park, Doocheon Seo, Heeseob Kim; GEO DATA. 2022;4(4):28-38. Published online December 31, 2022; DOI: https://doi.org/10.22761/DJ2022.4.4.004

1,671 View
97 Download

Abstract PDF: Korea Aerospace Research Institute (KARI) is conducting continuous quality control to provide reliable optical image products to various users. This paper describes KOrea Multi-Purpose SATellites (KOMPSAT-3 and KOMPSAT-3A) characteristics, operation, and image collection mode in order to enhance satellite image application. Also, image product of the satellites and quality management of the image product are described in this paper. The KOMPSAT-3 launched in 2012 and KOMPSAT-3A launched in 2015 collected many imageries around the world and provide them to users through web. Users can search for images through web catalog and order new imaging task. The KOMPSAT images provided under the KARI control is expected to be great help for earth observation and satellite image application enhancement.

Kompsat-5 Image Data Provision and Quality Management: Dochul Ynag, Horyung Jeong, Doochun Seo; GEO DATA. 2022;4(4):13-19. Published online December 31, 2022; DOI: https://doi.org/10.22761/DJ2022.4.4.002

2,058 View
138 Download

Abstract PDF: The Korea Aerospace Research Institute is conducting continuous quality management to provide reliable Kompsat-5 SAR image products to users. In this paper, the Kompsat-5 satellite operation, data processing, quality management, and data provision were described. The operation and image mode characteristics of the Kompsat-5 satellite from the image point of view were described, and the classification and characteristics of image products provided to users were explained. In addition, image data acquisition, quality index measurement, and its results are described for quality management of SAR images. Finally, it explains how to search for and order Kompsat image product through the ARIRANG system to quickly provide users with image products whose quality has been confirmed through quality management. Kompsat product can be searched and ordered from the ARIRANG Satellite Search and Order System (https://ksatdb.kari.re.kr/arirang/).

KOMPSAT-5 GNSS Radio Occultation Data Operations: Okchul Jung, Jaedong Seong, Myeongshin Lee, Daewon Chung; GEO DATA. 2022;4(3):1-7. Published online September 30, 2022; DOI: https://doi.org/10.22761/DJ2022.4.3.001

958 View
28 Download

Abstract PDF: The Korea Aerospace Research Institute launched KOMPSAT-5 on August 22, 2013, and has been operating for 10 years. KOMPSAT-5 has SAR (Synthetic Aperture Radar) for earth observation missions, and collects data necessary for earth atmosphere analysis through GNSS RO (Radio Occultation) receivers. RO data can be used for numerical weather forecast model based on temperature, pressure, and humidity by calculating the vertical distribution of atmospheric information. As a part of the Korea-US science and technology cooperation, KARI has been providing RO data of KOMPSAT-5 to the United States NOAA (National Oceanic and Atmospheric Administration) in near-real time since 2018. To this end, KARI receives telemetry data from the satellite about 12 times a day using 3 ground stations from Daejeon, Alaska in the U.S., and Sodankyla in Finland. The pre-processed data is being provided to both the UCAR (University Corporation for Atmospheric Research) in the U.S. and the KASI (Korea Astronomy and Space Science Institute). In this paper, radio occlusion data of KOMPSAT-5 is introduced, and system configuration, operation concepts for providing near-real time data and its application are also presented.

Radiometric Distortion Corrected Radar Backscattering Coefficient Data over Ilam, Iran using Kompsat-5 SAR Image: Dochul Yang; GEO DATA. 2021;3(4):28-31. Published online December 31, 2021; DOI: https://doi.org/10.22761/DJ2021.3.4.004

778 View
16 Download

Abstract PDF: Flattening gamma naught was calculated using Korea Multipurpose Satellite 5 (KOMPSAT-5, K5) by correcting the radiometric distortion caused by geometric distortion over Ilam, Iran. The flattening gamma naught is not only the SAR core observation of Analysis Ready Data (ARD), which is utilized for artificial intelligence and big data, but also the basis for all fields of application that use the SAR brightness by providing the backscattering values only from surface characteristics. The flattening gamma naught data is provided with the same resolution as that of the K5 SAR image, so the data over the Ilam, Iran have the spatial resolution of the K5 Wide Swath mode of 20 m. Shuttle Radar Topography Mission (SRTM) DEM with a resolution of 30 m was oversampled to generate the flattening gamma naught, and shadow areas where flattening gamma naught generation was not possible were identified using GIM layer information provided with the K5 image. In order to determine the reliability of the calculated flattening gamma naught, histogram analysis and tendency according to the incident angle were investigated, and the performance was verified by comparing it with other backscattering coefficients. Details of the algorithm and procedure are presented in previous studies and reference papers.

First
Prev
Page of 1
Next
Last

Search