tailings ponds detection

As an artificial structure, tailings ponds exhibit regular geometric shapes and relatively straight dams in HRRSIs. Because the typical tailings dam is composed of an initial dam and successive accumulation dams, the tailings dam structure presents obvious linear stripe characteristics. The initial dam, constructed using sand, gravel, or concrete, has a bright color, while the color of the accumulation dam varies based on factors such as particle size, soil coverage, and vegetation restoration. Collectively, these features can be leveraged to accurately identify tailings ponds in remote sensing images. Additionally, the exterior of the dam typically features various regular drainage pipes, ditches, roads, and vegetation, which create discernible point and line texture characteristics. These shapes, colors, textures, and associated infrastructures are pivotal for the accurate extraction of tailings ponds from HRRSIs.

Given the lack of a comprehensive registry for tailings dams, a dataset of tailings ponds was constructed, containing 3619 maps sourced from Google Earth. These maps reflect the actual dimensions of tailings ponds, which range from 2000 to 2500 m. [3]. The maps, featuring a resolution of 1.07 m, cover three channels (RGB). Labeling, a graphical image annotation tool written in Python, was used to label the samples, resulting in a total of 7624 annotations.