Nonlinear Thresholds Linking Street-Level Visual Exposure to Perceived Depressiveness in a High-Density University Town

High-density built environments raise concerns for mental health, yet most studies rely on macro-scale indicators and overlook the street-level visual environment that pedestrians experience. In dense campuses, depression among young adults is persistently high, but how micro-scale visual features shape perceived depressiveness through nonlinear pathways remains unclear. This study develops a spatially explicit perception–environment framework to examine these effects. Taking Longzihu University Town in Zhengzhou, China, as the study area, we combined large-scale street view imagery (SVI) with multi-source geospatial data: visual features were extracted by DeepLabV3+ semantic segmentation, and a Perceived Depressiveness Index (PDI) was derived from a pre-trained deep-perception model as the dependent variable. Sky View Factor (SVF) and Green View Index (GVI) were the dominant predictors. SVF followed a three-stage gradient in which perceived depressiveness declined continuously as sky openness increased, and GVI a non-monotonic pattern that was protective at low-to-moderate exposure and turned mildly adverse beyond a sign-change threshold of about 26%. Conditional interactions among GVI, SVF, Enclosure Index (ECL), road proportion, and sidewalk proportion revealed further threshold effects and diminishing returns. Perceived wellbeing is thus jointly shaped by micro-scale visual perception and macro-scale context, yielding quantitative thresholds to guide health-oriented campus greening and spatial planning. Methodologically, this study couples an interpretable XGBoost–SHAP framework with bootstrap-based threshold estimation and spatial validation to recover nonlinear thresholds and cross-feature interactions that linear or single-scale models cannot detect; empirically, sky openness (SVF) emerges as the dominant, direction-stable driver. Throughout, PDI denotes the perceived depressiveness of the street scene—an image-based environmental indicator—rather than clinically measured depression of individuals.