Integrating restorative perception into urban street planning: A framework using street view images, deep learning, and space syntax

Existing urban street planning often neglects human perception despite being designed for human utility. Therefore, exploring the relationship between urban street planning and human perception, especially restorative perception, is crucial for sustainable urban planning. In this study, we analyzed restorative perception of urban streets among different population groups and accessibility of urban streets using Baidu Street View Images, deep learning, and space syntax. Furthermore, we clarified the impacts of various street elements on restorative perception using correlation and ridge regression analyses. Based on restorative perception and accessibility coupling assessment, streets were classified into four types, identifying “inefficient segments” and “prioritized segments”. The results showed that: 1) the level of restorative perception of streets in Nanjing was relatively limited and exhibited variation across different population groups; 2) the impacts of different street elements on restorative perception varied, with trees having the highest positive impact (β = 0.378), while walls had the greatest negative impact (β = − 0.182); and 3) “prioritized segments” and “inefficient segments” represented 33.92 % and 17.96 %, respectively, indicating the urgent demand for streets planning and renewal. These findings can offer targeted recommendations for enhancing restorative environment of urban streets and identifying priority areas for urban street planning.