Non-Agricultural Change Detection in Multi-Scale and Multi-Season Remote Sensing Images

Non-agricultural change detection from remote sensing images plays an important role for food security. Currently, many change detection algorithms are encountering various difficulties and challenges when applied to remote sensing images of different scales and seasons. This paper proposed a multi-scale and multi-season dataset for non-agricultural change detection in remote sensing images, the dataset contains 500 pairs of high-resolution satellite images, covering samples with a resolution of 1m-2m and two diverse seasons, namely summer and winter. In the experiments, multiple state-of-the-art change detection models were utilized for validation of the dataset, in which BIT_CD achieved the highest F1 score of 80.62. The proposed dataset provides support for non-agricultural change detection of cultivated land.