An integrated deep learning approach for assessing the visual qualities of built environments utilizing street view images

Xukai Zhao, Yuxing Lu, Guangsi Lin

Abstract: Highlights•Advanced deep learning boosts accuracy in evaluating visual aspects of built environments for authorities.•Strong correlation found between the spatial distribution of objective visual elements and subjective perceptions.•SHapley Additive Explanation (SHAP) visualizations underscore street view elements' impact on subjective perceptions.•Class Activation Map (CAM) visualizations align with eye-tracking study findings, offering new insights.