A Study on Urban Platycladus orientalis Species Identification Using an Enhanced YOLO Model and High-Resolution Satellite Imagery

The conflict between public health risks caused by allergenic tree species in urban greening and the goal of promoting urban ecological biodiversity has become a major challenge for allocating urban ecological resources. In March 2025, the “yellow haze incident” in the Temple of Heaven area in Beijing intensified this conflict and highlighted the problem of pollen allergenicity in urban plants, drawing wide attention to highly allergenic tree species used in urban landscaping. This article proposes a tree species identification method that combines high-resolution satellite imagery with the YOLOv8s object detection model and focuses on Platycladus orientalis. Three lightweight modules, the spatial context-aware module, the feature reconstruction module, and the bidirectional feature pyramid network, are added to the neck of the baseline model to enhance its ability to identify Platycladus orientalis in remote sensing images. The enhanced YOLOv8s model achieves a precision of 81.39%, a recall of 78.70%, an mAP50 of 83.32%, and an F1 score of 80.81% for urban Platycladus orientalis detection. Using this model, we extract 118 723 Platycladus orientalis instances from submeter satellite remote sensing imagery of Beijing and generate a spatial distribution map. These results show that the enhanced YOLOv8 model can support the identification and analysis of the spatial distribution of the target tree species. The proposed method provides basic data and decision support for urban plant species identification, statistical analysis, and planning management.