Sidewalk extraction using aerial and street view images

A reliable, punctual, and spatially accurate dataset of sidewalks is vital for identifying where improvements can be made upon urban environment to enhance multi-modal accessibility, social cohesion, residents' physical activity. This paper develops a synthetically new spatial procedure extract the sidewalk by integrating detected results from aerial street view imagery. We first train neural networks images, then use pre-trained models restore occluded missing images. By combining both data sources, complete network produced. Our case study includes four counties in U.S., precision recall reach about 0.9. The imagery helps largely enhances network's connectivity linking 20% dangles.