Abstract: Fine-scale characterization of urban tree resources is of great significance for deepening the cognition of urban green space systems and improving the refined management level of urban green spaces. Taking Yaohai District, Hefei City, Anhui Province as the research object, this paper constructs an urban individual tree identification and canopy segmentation method based on the HR-SFANet model using Google remote sensing imagery with a 1 m resolution, and realizes high-precision extraction of the spatial distribution and structural information of trees in the study area. The results show that the overall accuracy of this model reaches 77.20%, with favorable individual tree identification performance and stability. In this study, approximately 182,300 individual trees were identified in Yaohai District, Hefei City, with an average tree density of about 7.35 trees/hm². The overall spatial pattern presents a feature of ‘high quantity in the peripheral areas and high density in the central areas'. Differences in community types, land use structure and built environment are important factors leading to the spatial differentiation of tree resources. On this basis, this paper analyzes the tree allocation characteristics of different types of communities, and puts forward refined management strategies such as identification of priority replanting areas, hierarchical maintenance management and green space connectivity optimization. The research results provide technical support and decision-making reference for the investigation, dynamic monitoring and refined governance of urban green space resources.