E-Unet++: A Semantic Segmentation Method for Remote Sensing Images

Abstract

Semantic segmentation can distinguish objects in remote sensing images at the pixel level. However, traditional semantic segmentation algorithms are more and more difficult to meet people's needs. With the rapid development of deep learning, especially its application in remote sensing images has greatly improved the parsing ability and efficiency. But, the complexity and diversity of remote sensing image content make the accuracy of semantic segmentation still need to be improved. Thus, a semantic segmentation method that combines the characteristics of EfficientNet and UNet++ is proposed in this paper. The method can make the segmentation boundary clearer and improve the segmentation effect of densely distributed objects. The results show that the proposed method achieves good performance in the Vaihingen dataset.