論文ID: 2021-B023
Recent studies demonstrated associations between physical environment (especially greenery) and people’s health, well-being, and crime rate by using street-level imagery as ‘big data’ and automated image recognition methods. However, few prior studies focused on interrelations between physical environment and residents’ social relationships. This study investigated associations between physical environments and psychological tendencies of neighboring communities in Japan by using a mail survey and Google Street View images. The mail survey was collected from 156 regions across eight prefectures in western Japan. Google Street View images of these regions were collected and classified by machine learning models and human observers. The results indicated mainly negative correlations between the survey items related to feelings towards participants’ neighbors such as social capital and the rate of outdoor gardening by region. Additionally, these correlation patterns differed by type of community, namely, fishing, farming, and other types of communities.
