Farmland Flood Area Segmentation Dataset

The current agricultural sector faces challenges brought by frequent flood attacks, especially in farmland management and crop protection. Existing flood monitoring solutions often rely on traditional remote sensing technology, which lacks timeliness and accuracy. Therefore, there is an urgent need for an efficient dataset to support AI model training to achieve precise identification of flooded boundaries and ranges. This dataset provides pixel-level segmentation annotations of flood-covered farmland areas, aiming to enhance AI model recognition capabilities in this scenario. Data collection is mainly conducted via drones equipped with high-definition cameras during flood events to capture real farmland flood conditions. Additionally, multiple rounds of annotation and expert review quality control measures are taken to ensure the accuracy and consistency of the data annotations. Data is stored in PNG/JPG format, organized in a way that each image corresponds to its segmentation mask, facilitating subsequent model training and evaluation.