Revisiting Gehl’s urban design principles with computer vision and webcam data: Associations between public space and public life

Abstract

Understanding pedestrian behavior is crucial to inform public space design. However, being laborious, Gehl’s Public Space and Public Life (PSPL) framework is restricted to a small scale. Although prior studies have utilized computer vision (CV), they either focused on monitoring social distancing or measuring urban vitality, ignoring the subtle interplay between public space and public life. This study utilizes webcam data to track walk and stay behaviors, investigating their associations with public space features including point of interest, façade quality, and street furniture. Our findings extend PSPL principles. First, pedestrians tend to stand in less private places with good visual connectivity, indicating that privacy matters less to standing than sitting. Second, pedestrians walk along the edge in large-scale spaces while keeping in the middle in small spaces. Third, although all POIs affect vitality, certain types are more effective (i.e., catering). Fourth, a good place to stay must be convenient to walk through. Our CV framework partially automates PSPL without incurring labor costs. Urban design studies can use the operationalized CV pipeline to draw evidence-based design recommendations and monitor people-space interactions at large scale.