The paper introduces a deep learning-based image segmentation algorithm. The algorithm builds a local semantic information and global semantic information fusion model to improve the segmentation accuracy and reduce the algorithm's computational complexity.
