The automatic extraction of buildings from highresolution aerial imagery plays a significant role in many urban applications. Recently, convolution neural network (CNN) has gained much attention in remote sensing field and achieved remarkable performance in building segmentation from visible aerial images. However, most of the existing CNN-based methods still have the problem of tending to produce predictions with poor boundaries. To address this problem, in this paper a novel semantic segmentation neural network named Edge-Detail-Network (E-D-Net) is proposed for building segmentation from visible aerial images. The proposed E-D-Net consists of two subnetworks E-Net and D-Net. On the one hand, E-Net is designed to capture and preserve the edge information of the images. On the other hand, D-Net is designed to refine the results of E-Net and get a prediction with higher detail quality. Furthermore, a novel fusion strategy which combines the outputs of the two sub-networks is proposed to integrate edge information with fine details. Experimental results on the Inria Aerial Image Labeling Dataset and the ISPRS Vaihingen 2D semantic labeling dataset demonstrate that, compared with the existing CNN-based model, the proposed E-D-Net provides noticeably more robust and higher building extraction performance, thus making it a useful tool for practical application scenarios.
Learning from a limited number of samples is challenging since the learned model can easily become overfitted based on the biased distribution formed by only a few training examples. In this paper, we design a novel few-shot learning method to solve the problem of tire pattern classification proposed by Xi'an University of Posts and Telecommunications laboratory. The proposed method consists of two steps. On the one hand, we calibrate the distribution of these fewsample classes by transferring statistics from the classes with sufficient features (FD). On the other hand, an adequate number of examples can be sampled from the feature distribution to expand the inputs to the classifier (SVM). Experimental results on the Tire pattern dataset demonstrate that, compared with the existing few-shot learning models, the proposed FT with SVM provides noticeably more robust and higher performance, thus making it a useful tool for practical application scenarios.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.