Objective As one of the major open spaces, urban streets not only carry the transportation function, but also promote social interaction and economic growth, while displaying the aesthetic, cultural and historical characteristics of the city. However, many new or regenrated streets have gradually lost their identity due to the compromise of large-scale construction and short-term delivery. It is important to explore ways to protect and optimize the locality of streetscape since street constituents the most direct impression of a city, which not only forms the identity of a city together with architecture, historical and cultural heritage, geographical characteristics and social life, but also serves as an independent carrier of local expression. The identification and optimization of streetscape locality can help protect the uniqueness of landscape, enhance urban identity and construct urban image, especially for historical and cultural preservation areas in modern cities, and can also work as an important step to retain urban memory and enhance residents’ sense of belonging.

Methods This research first establishes a theoretic framework for streetscape locality through a discussion based on literature reviewed. The research proposes that streetscape locality should include the physical, social, historical and aesthetic aspects, with relevant indicators being identified for the four aspects respectively. These indicators are then compared with streetscape characteristics that can be measured by current image analysis technologies. On this basis, a framework for identification of streetscape locality is constructed. Taking Shanghai Hengfu Historic and Cultural Preservation Area (hereinafter referred to as the “Area”) as an example, the research adopts the semantic segmentation method and an AI-based evaluation model to analyze the locality of the Area, with the specific process being visualized using the Geographic Information System (GIS). The current spatial expression of streetscape locality characteristics in the Area is then discussed considering factors such as land use on both sides of a street, architectural styles, and traffic hierarchies.

Results Results suggest that in the aspect of physical locality, the indicators of enclosing degree, green visibility and sky visibility are consistent in spatial distribution. Specifically, the distribution of sky visibility is basically opposite to green visibility, while enclosing degree is similarly distributed with sky visibility. In general, the enclosing degree of the Area is rated moderate, while the green visibility and sky visibility of streetscape are rated at a medium to low level. Streetscape in the northern and western parts of the Area is relatively open compared with other parts and the highest green visibility is observed in streets around Xujiahui Park located in the southwestern part of the Area. In the social aspect, safety is rated good overall, but the human scale is rated moderate and the imageability level is low. Streetscape with appropriate human scale is safer than others, which may be due to the fact that both the two human scale and safety characteristics can positively influence the comfort of perception. Overall, safety and human scale characteristics are poor in the central part of the Area around Shanghai Conservatory of Music, the intersection betwwen Huaihai Middle Road and Fuxing Middle Road, as well as Yan’an Middle Road in the north and Chongqing South Road in the east of the Area. Besides, the streetscape imageability of the Area is only rated high where unique landmarks exist. In the historic aspect, legibility of the Area appears to be better in the southern part while rated poorest in the central part and western boundary of the Area, especially in Jiangsu Road and Nanchang Road close to the north side of the Fuxing Park. Among the characteristics in relation to aesthetic aspect, visual diversity, coherence and permeability show no direct or indirect correlation with each other in spatial distribution. The visual diversity and coherence of streetscape are rated good in the research area, but the permeability of streetscape is less satisfied. The visual diversity of streetscape at the intersection between Changshu Road and Huaihai Middle Road and along Jianguo West Road in the south is poor, and the permeability is also quite poor. In addition, the roads with low traffic hierarchy such as branches and alleys are generally rated better than those with high traffic hierarchy in terms of human scale, coherence and permeability.

Conclusion This research may provide a reference for subsequent protection, optimization and development of urban streetscape locality, and may also inspire more large-scale analysis of landscape concepts derived from sociology, anthropology and other disciplines in the future. Constrained by the accuracy of the algorithm model adopted and the availability of the streetscape image data collected, there may exist slight error in the analysis results. However, the research in general has succeeded in quantitatively measuring and mapping the locality of streetscape within the research area.