Enhanced Automatic Identification of Urban Community Green Space Based on Semantic Segmentation

At the neighborhood scale, recognizing urban community green space (UCGS) is important for residential living condition assessment and planning. However, current studies have embodied two key issues. Firstly, existing focused on large geographic scales, mixing rural areas, neglecting accuracy of contours at fine scales. Secondly, spaces covered by shadows often suffer misclassification. To address these issues, we created a neighborhood-scale dataset proposed segmentation decoder HRNet backbone with auxiliary decoders. Our model adds additional branches to low-resolution representations improve their discriminative ability, thus enhancing overall performance when high- are fused. evaluate model, tested it that includes satellite images Shanghai, China. The outperformed other nine models in UCGS extraction, precision 83.01, recall 85.69, IoU 72.91, F1-score 84.33, OA 89.31. also improved integrity identification shaded over HRNetV2. method could offer useful tool efficient detection mapping