A BOUNDARY AWARE NEURAL NETWORK FOR ROAD EXTRACTION FROM HIGH-RESOLUTION REMOTE SENSING IMAGERY
Keywords: Road Extraction, Deep Learning, Semantic Segmentation, Coarse to Fine Learning, Boundary-aware, Boundary Quality
Abstract. Automatic road extraction from high-resolution remote sensing imagery has various applications like urban planning and automatic navigation. Existing methods for automatic road extraction however, focus on regional accuracy but not on the boundary quality. To address this problem, a Boundary-aware Road extraction Network (BARoadNet) is proposed. BARoadNet is a coarse-to-fine architecture composed of two encoder-to-decoder networks, a Coarse Map Predicting Module (CMPM) and Fine Map Predicting Module (FMPM). The CMPM learns to predict coarse road segmentation maps. The FMPM is used to refine the coarse road maps by learning the difference between the coarse road extraction result and the ground truth. Experiments are conducted on the open Massachusetts Road Dataset. Quantitative and qualitative analysis demonstrate that the proposed BARoadNet can improve the quality and accuracy of road extraction results compared with the state-of-the-art methods.