Exploring the impact of street-level built environment characteristics on vehicle speeds in low-speed streets: A case study of Tainan city

This study investigates the impact of street-level built environment characteristics on vehicle speeds in Tainan's low-speed urban streets. Employing a multi-source data integration approach, we combined Google Directions API, OpenStreetMap, and Street View Imagery. Semantic segmentation quantified visual proportions of environmental elements as proxies for roadside friction, while POI (Point of Interest) data indicated functional density. Multiple linear regression identified key determinants of speed. Results show that higher visual proportions of buildings and sidewalks significantly correlate with reduced vehicle speeds, reflecting a “roadside friction” effect that prompts cautious driving. Conversely, residential segments exhibited a positive correlation with speed, suggesting a lack of perceived visual constraints. By bridging computer vision and urban analysis, this study provides a scalable framework for human-centered transportation planning. The findings offer practical insights for designing effective traffic-calming measures in dense Asian cities, emphasizing the role of perceived environmental friction in governing driver behavior.