Mapping city-wide allergenic risk zones of Platanus trees using street view images and deep learning

Urban trees provide essential ecological and societal benefits; however, certain species such as Platanus x hispanica; Mill. ex Münchh (London plane) releases allergenic pollen that poses health risks to sensitive populations. Existing studies on allergen exposure often rely on sparse monitoring stations, limiting the application in generating city-wide allergenic risk map. This study proposes an effective and low-cost framework for mapping urban allergenic risk by integrating Street View Images (SVIs), deep learning, and environmental modeling. Using over 150,000 SVIs from Wuhan, Platanus trees are automatically identified through a fine-tuned YOLOv11 (You Only Look Once) instance segmentation model. A Platanus Displayed Density (PDD) indicator is proposed by integrating monocular depth information to measure visually perceived Platanus abundance from a pedestrian exposure perspective. Building upon this, a novel Allergenic Risk Index of Platanus (ARIP) is developed by combining tree abundance, urban morphological effects (building hindrance index), and wind-driven pollen dispersion model. In the result, we also present city-wide maps with high-resolution (100 m) of allergenic risk and population exposure using mobile phone data. The proposed framework provides an innovative, transferable, and data-efficient approach for fine-scale allergenic risk assessment, supporting allergy-aware urban planning and public health management in cities.