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

Abstract

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

Localizing and prioritizing roof greening opportunities for urban heat island mitigation: insights from the city of Krefeld, Germany

“…machine learning approaches (Shao et al 2021). For example, Chen et al (2022) proposed a segmentation decoder for the High-Resolution Network (HRNet) model that accurately detects urban green on a fine scale with improved performance on shaded green.…”

Quantifying greenspace using deep learning in Karachi, Pakistan

“…We apply DeepLabV3+, a deep convolutional neural network based on parallel atrous convolution structures [11]. This model has been shown to achieve state-of-the-art semantic segmentation performance for vegetation detection and classification [3,2,10,13]. In contrast to vegetation indices, the deep learning approach can distinguish greenspace classes, thus here is also trained to identify the greenspace types, tree and grass.…”

“…This problem of class scarcity hinders deep neural network training, which relies on large volumes of data to extract representative landscape features [47,58]. Though studies have adopted image augmentation techniques including random flips, rotations, brightness and contrast change to tackle class scarcity [24,10,53], such techniques only increase data variety with respect to basic image properties, and do not provide additional information for the important target feature, greenspace.…”

Research on Post-Use Evaluation of Community Green Space Rectification Based on a Multi-Dimensional Perception System: A Case Study of Jiayuan Sanli Community in Beijing

“…Thanks to the increasing sophistication of deep learning algorithms, evaluating data and outputting planning guidelines through AI models is a reality [26]. Introducing deep learning algorithms into a post-use evaluation system can be an effective way for planners to make subjective judgements on the output data of the coupled system [27].…”