AIWR: Aerial Image Water Resource Dataset
Description
Aerial Image Water Resource (AIWR) Dataset According to the land use code standards set by the Fundamental Geographic Data Set (FGDS), land use classification in Thailand requires the analysis and transformation of satellite imagery in conjunction with field survey data. In this study, researchers focused specifically on land use in water bodies. The water bodies are classified into two categories: natural bodies of water (W1) and artificial bodies of water (W2). The aerial images used in this research were captured at a resolution of 1:50 meters, with each image measuring 650x650 pixels. These images included both types of water bodies, W1 and W2. Ground truth annotations were established for all aerial images prior to analysis by remote sensing experts to ensure accurate classification of the water bodies. An example of ground truth, validated by experts, was used to train the deep learning algorithm and further assess its performance. The AIWR dataset, which contains 800 aerial images, consists of both W1 and W2 water bodies. The data were randomly divided into two sets: training and test sets, with a 9:1 ratio. Consequently, 720 aerial images were used for model training and development, while 80 images were reserved for testing.
Files
Institutions
Categories
Funding
Mahasarakham University