The built environment characteristics associated with walkability range from neighborhood-level urban form factors to street-level urban design factors. However, many existing walkability indices are based on neighborhood-level factors and lack consideration for street-level factors. Arguably, this omission is due to the lack of a scalable way to measure them. This paper uses computer vision to quantify street-level factors from street view images in Atlanta, Georgia, USA. Correlation analysis shows that some streetscape factors are highly correlated with neighborhood-level factors. Binary logistic regressions indicate that the streetscape factors can significantly contribute to explaining walking mode choice and that streetscape factors can have a greater association with walking mode choice than neighborhood-level factors. A potential explanation for the result is that the image-based streetscape factors may perform as proxies for some macroscale factors while representing the pedestrian experience as seen from eye-level.

与步行性相关的建筑环境特征从邻里级城市形态因素到街道级城市设计因素不等。但是，许多现有的步行能力指标都是基于邻里级别的因素，而没有考虑街道级别的因素。可以说，这种遗漏是由于缺乏可扩展的方式来测量它们。本文使用计算机视觉从美国佐治亚州亚特兰大市的街景图像中量化街道级因素。相关分析表明，某些街景因素与邻里级因素高度相关。二元逻辑回归表明，街景因素可以显着有助于解释步行模式选择，并且街景因素与步行模式选择的关联性要比邻里级因素更大。