A Spatial Semantics and Continuity Perception Attention for Remote Sensing Water Body Change Detection (ArXiv)
The HSRW-CD dataset is the first high spatial resolution dataset dedicated to WBCD (Water Body Change Detection). It comprises 2,085 image pairs spanning diverse geographical regions and environmental contexts, encompassing various water body types such as urban waterways, river systems, lacustrine environments, and artificial reservoirs. These image pairs are from more than 7 cities or districts in China, including Beijing, Chongqing, Chengdu, Hangzhou, Wuhan, Shenzhen, and Shanghai. The bi-temporal images are manually collected from high-resolution satellites and Google Earth. The annotation of the dataset is carried out by an expert group of Earth vision applications, which guarantees high label accuracy. In addition, the bi-temporal imaages have a high spatial resolution finer than 3 meters, each sized at 512 × 512 pixels. Following a 7:1:2 proportion, the image dataset is randomly divided into training, validation, and test subsets, leading to training, validation, and test sets containing 1,476, 203, and 406 independent image pairs.
Obtain it by: [Baidu Netdisk-Link] (https://pan.baidu.com/s/1wygwa15uOreD3-z_MT3wPw?pwd=opac)
Obtain these by: [Quark Netdisk-Link] (https://pan.quark.cn/s/476e779db6d1?pwd=huyB)