With the emergence of human-centered urban development goals and the increasing pursuit of a better quality of life, the architectural façades of cities are receiving growing attention. However, during the process of urban development, architectural façades often experience physical disorder. This phenomenon tends to be overlooked in targeted urban management practices or lacks cohesive urban renewal planning at a macro scale. This oversight can negatively impact the livability and attractiveness of a region. This study aims to quantify the architectural façades encountered daily by urban residents by measuring the physical disorder of architectural façades to inform better urban renewal using deep learning and space syntax. First, streetscape images of architectural façades were collected using the Baidu Maps Street View service. Subsequently, an evaluation system for architectural façades was developed, and machine learning was employed to conduct high-resolution measurements and assessments of these façades. Simultaneously, street network data is extracted and analyzed using space syntax to quantify the accessibility of architecture on each street. Finally, by integrating the analysis of architectural façades and accessibility, the study identifies priority areas for building renewal, thus providing a decision-support tool for sustainable urban renewal planning. Overall, the paper presents an innovative method that combines image data, deep learning, and space syntax-derived architectural accessibility for a quadrant analysis. It offers designers and decision makers new perspectives and enhances the livability of residents by focusing on the physical condition of architectural façades, thereby making urban renewal practices more human-centered and better aligned with the actual needs of city dwellers.
Read full abstract