Street Sky Openness Estimation Based on Deep Learning——Case Study of Hakata Station Area, Fukuoka

Abstract

The sky view factor has been recognized as an important indicator in the urban landscape and it is closely related to the physical and mental health of urban residents. In the past studies, the calculation of sky openness and the calculation of measurement indices were still mostly done manually, which was time-consuming and laborious. Meanwhile, Walk Score is widely used to measure the accessibility of surrounding amenities, often considering the physical characteristics of streets. In this study, we provided a method to calculate the sky openness of street-based on semantic segmentation processing of street view images. Based on a deep learning model, this study semantically segmented the streetscape images of roads around the Hakata station area in Fukuoka, Japan. The results could be compared with the Walk Score. The correlation analysis between the street sky openness and Walk Score would also be conducted. The findings showed that there is a negative correlation between street sky openness and Walk Score. This study is based on the estimation of street view images; therefore, it can more truly reflect the current situation of the street, and the model's high efficiency also makes it more suitable for large-scale urban research.