I. Introduction
In recent years, with the wide application of AI in agriculture, smart agriculture has become a major research direction for modern and future agricultural development. The appearance of remote sensing by UAV is furthering the development of smart agriculture [1]. Farmland segmentation based on remote sensing images is an important research direction for smart agriculture and an important foundation of smart farmland management. Therefore, the study of an accurate farmland segmentation method that can be applied to high-spatial-resolution remote sensing images is important for the development of smart agriculture.