"FDTNet dataset"

"The WHU Building Change Detection Dataset (WHU-CD) is a benchmark dataset for building change detection in high-resolution aerial imagery. The dataset covers the city of Christchurch, New Zealand, which experienced a magnitude 6.3 earthquake in February 2011. It consists of a pair of bitemporal aerial images acquired before the earthquake (2012) and after reconstruction (2016) over the same geographic area. The spatial resolution is 0.2 meters per pixel, and the full image dimensions are 32,507 \u00d7 15,354 pixels. The dataset provides pixel-level binary change labels distinguishing changed buildings from unchanged areas. For convenience, the original large images are cropped into 1,260 training patches and 690 testing patches, each of size 256 \u00d7 256 pixels. WHU-CD has been widely adopted as a standard benchmark for evaluating building change detection methods, particularly for large-scale, high-resolution remote sensing applications. The original dataset is publicly available at https:\/\/study.rsgis.whu.edu.cn\/pages\/download\/building_dataset.html.The LEVIR-CD (LEarning, Vision, and Remote sensing - Change Detection) dataset is a large-scale benchmark dataset for building change detection in remote sensing imagery. The dataset contains 637 pairs of bitemporal Google Earth images collected from 20 different urban regions across Texas, United States, spanning the years 2002 to 2020. Each image has a spatial resolution of 0.5 meters per pixel and dimensions of 1,024 \u00d7 1,024 pixels. The dataset provides pixel-level binary change masks with over 31,000 independently labeled building change instances. LEVIR-CD is significantly larger than previous change detection datasets (two orders of magnitude), offering a challenging benchmark with diverse building appearances, seasonal variations, shadows, and lighting conditions. It has become one of the most widely used datasets for evaluating deep learning-based change detection methods. The original dataset is publicly available at https:\/\/justchenhao.github.io\/LEVIR."