Abstract
Urban green spaces (UGSs) provide essential environmental services for the well-being of ecosystems and society. Due to the constant environmental, social, and economic transformations of cities, UGSs pose new challenges for management, particularly in fast-growing metropolitan areas. With technological advancement and the evolution of deep learning, it is possible to optimize the acquisition of UGS inventories through the detection of geometric patterns present in satellite imagery. This research evaluates two deep learning model techniques for semantic segmentation of UGS polygons with the use of different convolutional neural network encoders on the U-Net architecture and very high resolution (VHR) imagery to obtain updated information on UGS polygons at the metropolitan area level. The best model yielded a Dice coefficient of 0.57, IoU of 0.75, recall of 0.80, and kappa coefficient of 0.94 with an overall accuracy of 0.97, which reflects a reliable performance of the network in detecting patterns that make up the varied geometry of UGSs. A complete database of UGS polygons was quantified and categorized by types with location and delimited by municipality, allowing for the standardization of the information at the metropolitan level, which will be useful for comparative analysis with a homogenized and updated database. This is of particular interest to urban planners and UGS decision-makers.
Highlights
Urban green spaces (UGSs) face significant challenges due to rapid urbanization and climate change [1]
The results indicate that for over 15 years, the overall UGSs were reduced to 50%
Because the input data consisted of three-band composed images, the Dice coefficient was computed for each class and averaged via arithmetic mean through the fastai implementation [77]
Summary
Urban green spaces (UGSs) face significant challenges due to rapid urbanization and climate change [1]. This process first reduces the size and increases the number of bands of the training images and their activation maps generated in each layer of the network to subsequently carry out the opposite process considering information from the encoder in the segmentation of fine details [44] These types of networks have achieved wide success with state-of-the-art results for a wide variety of problems from medical applications [45,46] to their employment in remote sensing for road [47] and building extractions [48], as well as land cover classification [49], but they have not been used to make many advances in the UGS area. The process involves different convolutional neural network encoders on the U-Net architecture with the use of threeband compositions of very high resolution (VHR) satellite imagery channels and vegetation indices as input data This precise and updated data collection and new UGS cartography at the metropolitan level would improve the understanding of connectivity and accessibility of UGSs as a basis for management and decision-making for land use in urban areas
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
