Smart Assessment and Improvement of Urban Street Spaces: A Case Study in Binjiang, Hangzhou

This study aims to build a quantitative assessment framework of urban spatial quality through deep learning technology and multi-source data fusion. However, it is important to acknowledge that the data from Baidu Maps API and OpenStreetMap may have limitations due to the frequency of updates and coverage of these platforms. To mitigate these limitations, the research team cross-referenced the collected data with additional sources and conducted field surveys to ensure a more accurate representation of the urban environment. The pixel-level semantic segmentation technique based on the FCN network model was used to intelligently recognise and quantitatively analyse the urban streetscape. The results reveal the current status of key indicators such as the level of greening, sky openness, building enclosure, and the density and mix of POI facilities in the streetscape of Binjiang District, and propose targeted optimisation suggestions to improve the quality of the urban space and meet residents' needs for a better living environment.