Does streetscape color matter for urban perceptions? A deep learning approach to street view images

Abstract

Color influences people's visual perception, making it an essential component of both the natural and built environments. As a result, more thoughtful land use planning is needed. In this vein, this study re-examined the association of streetscape color features and urban perceptions at a city wide. We innovatively applied an AI approach on street view images and deep learning to measure six streetscape color features and various urban perceptions. Our results demonstrate that streetscape color feature have significant impact on urban perceptions, while color compatibility is not always as critical as people have long assumed with no significant impact on wealth, safety, or comprehensive perception. We found a consistently positive effect of street color similarity, bilateral color similarity, and color balance on all dimensions of urban perception. And the impact of aesthetically related features varies across different perceptions, including colorfulness, color diversity and color compatibility.