Urban Street Space in Planning Transformation: Analyzing and Optimizing the Quality Differences between Subjective and Objective Perspectives—A Case Study of Xi’an

Abstract

As the most common public spaces, urban streets are not only vital for daily activities but also a direct reflection of the quality of the urban environment and the inclusiveness of society. In the face of multiple challenges such as global climate change, resource crises, and social inequality, enhancing the spatial quality of urban streets has become crucial for improving environmental sustainability, social inclusiveness, and livability. However, most existing research focuses on either objective environmental features or subjective perceptions, often overlooking the systematic differences between these two aspects and their profound impact on spatial design and planning decisions. In the context of rapid urbanization, how to scientifically and quantitatively identify the differences between subjective and objective quality and develop targeted optimization strategies remains a critical issue that needs urgent attention. In this study, an innovative spatial quality evaluation framework based on multi-label classification is proposed, integrating deep learning technology and semantic analysis methods. Using the MIT Place Pulse dataset, which includes 110,988 street view images, the Microsoft TrueSkill algorithm quantifies six subjective perception scores for streets, including safety, comfort, and richness. Additionally, using the Cityscapes dataset and the Mask2Former algorithm, the objective features of 28,662 streets in the central area of Xi ‘an are semantically segmented, and eight landscape element indicators are extracted from four dimensions: safety, comfort, richness, and convenience. The study reveals that (1) the spatial quality of the main urban area of Xi’ an exhibits significant spatial heterogeneity and stratification in both subjective perception and objective physical environment dimensions, with higher scores in the core urban area compared to the peripheral areas. (2) There are significant differences in subjective and objective evaluations across different regions, with weak overall correlation between the two, and some administrative districts even showing statistically significant negative correlation. (3) Most streets exhibit inconsistencies between subjective and objective evaluations, particularly those where ‘subjective is better than objective’ and ‘objective is better than subjective,’ with a low proportion of streets achieving high consistency in both subjective and objective evaluations. Therefore, improving the quality of urban streets requires not only focusing on the construction of ‘hard environments’ but also emphasizing the coordinated optimization of subjective experience and social environment. Through quantitative analysis, this study reveals the key deviations and optimization paths of street space from subjective and objective perspectives, which provides practical basis and theoretical enlightenment for street planning and design under the background of rapid urbanization, and helps to build more inclusive and livable urban public space.