Evaluating building color harmoniousness in a historic district intelligently: An algorithm-driven approach using street-view images

Building color and harmoniousness have been regarded as critical issues in planning historic districts. Harmoniousness of building façade colors (HBFC) is an indicator to evaluate the quality of the built environment, which can be perceived but is difficult to measure quantitatively. In addition, alleviating the impact of shadows in street-view images (SVIs) to assess building façade color is another research gap that is difficult to address. This paper proposes an efficient approach for evaluating HBFC on a large-scale using SVIs and a deep learning algorithm. Specifically, a shadow processing method was developed, and transfer learning was integrated into the harmoniousness evaluation process. The historical district of Guangzhou, China, was selected as a case study area. This study contributes to the development of human-centered planning and design by providing continuous measurements of “unmeasurable” quality across large-scale areas. Meanwhile, insights into building façade color and its harmoniousness can assist with accurate design guidance, which is important for historic districts.