A Fuzzy-Boundary Enhanced Trident Network for Parcel Extraction in the Urban–Rural Area

As the basic unit of farmland, parcel is crucial for remote sensing tasks, such as urban management. Previous studies farmland parcels extraction are based on boundary detection and instance segmentation methods. However, these methods perform poorly in with complex shape fuzzy due to insufficient feature capability. Moreover, lack multi-scale features fusion, they difficult extract different scale accurately. Based issues, we propose a Fuzzy-Boundary Enhanced Trident Network, named FBETNet, enhance generate parcels. First, semantic-guided multi-task strategy introduced order boundary. Second, design trident module further improve performance extraction. Finally, adversarial data augmentation employed training phase strengthen robustness stability out proposed method. Experiments show that our method improves significantly both accuracy visualization, especially shape.