关键词: 航空图像, 深度学习, 语义分割, 损失函数, 边界约束

Abstract: The advancements in deep learning and computer vision technologies have had a profound impact on the field of air-borne remote sensing, making the analysis of aerial images more efficient. Compared to conventional images, the target boundaries in aerial images are clearer and more distinct, with more regular distributions and stronger spatial structure. However, current state-of-the-art segmentation methods mainly focus on utilizing complex feature extractors to capture stronger contextual relationships, placing more emphasis on single-pixel classification accuracy. This not only demands higher hardware requirements but also overlooks the issue of boundary alignment from a structural perspective. To ad-dress this challenge, we propose an innovative boundary-aware loss function, Lossd, designed to enhance the perfor-mance of semantic segmentation for aerial remote sensing images, particularly in terms of boundary precision and target segmentation consistency. We innovatively translate structural differences into a loss, unlike traditional methods that fo-cus on single-pixel accuracy. Moreover, we propose an effective solution for the common over-segmentation and under-segmentation problems in semantic segmentation tasks. Extensive experimental validation has been conducted on three widely used large-scale datasets and three benchmark models. Experimental results show that our method significantly improves the semantic segmentation performance of the model without modifying the original network. On the LoveDA dataset, it achieves 55.8% mIoU (+1.6%), on UAVid, 70.8% mIoU (+0.8%), and on Potsdam, 94.1% mF1 (+0.7%), ap-proaching and partially surpassing state of the art. We will open-source our code at https://github.com/qwqwqer233/Lossd.

Key words: Aerial Images, Deep Learning, Semantic Segmentation, Loss Function, Boundary Constraints