Objective Walking is the most fundamental mode of urban transportation, and building pedestrian-friendly streets holds great significance in urban planning. Yet many metropolises remain auto-oriented and lack high-quality pedestrian environments. Advances in multi-source urban data and deep learning now make it feasible to evaluate street-level walkability with greater coverage, precision, and reproducibility than traditional audit or perception-based approaches. This study proposes a comprehensive framework that integrates multi-source urban data with deep learning to quantify urban street walkability, and demonstrates its utility through an application to Beijing’s central districts with external validation against on-site pedestrian ratings.

Methods By reviewing relevant literature, a multi-criteria street walkability value framework was developed, covering five dimensions: imageability, convenience, vibrancy, comfort, and safety. A walkability evaluation indicator system comprising 23 indicators was established. To operationalize the measurement, we assembled and harmonized multi-source urban data: vector layers for roads, buildings, green spaces, and land-use types; high-resolution satellite imagery; 213,950 street-view images captured via the Baidu panorama application programming interface (API) at 100-meter intervals and four bearings per point; point of interest (POI) records (20 top-level categories); urban heat-map rasters sampled at six time slots across a workday−weekend cycle; field photos; and official statistics. The street network was segmented into a 100 m grid of 53,631 units after removing duplicates and very short fragments. A DeepLab v3+ semantic segmentation model, fine-tuned via transfer learning on urban streetscape data and optimized with cross-validation and learning-rate decay, produced pixel-level shares of sky, vegetation, roadway, buildings, and other salient elements for each viewing direction; a weighted fusion yielded panoramic proportions per street unit. A YOLO v5 detector, trained on 2,000 labeled images based on building-facade quality criteria and validated on a 10% hold-out set, identified facade attributes relevant to building quality. ArcGIS pipelines performed accessibility analysis, conducted geo-joins of POIs and heat-map intensities to the street grid, integrated building and population data through grid-based analysis, and visualized the final results; georeferencing of heat-map mosaics used a WGS-84 frame and second-order polynomial transformation. Indicator weights were derived using the analytic hierarchy process (AHP) based on judgments from 35 domain experts (consistency ratio CR < 0.10). Indicator values were min-max normalized and combined via weighted summation to obtain a composite walkability score for each street unit.The empirical application covers Beijing’s central urban districts (Dongcheng, Xicheng, Chaoyang, Haidian, Fengtai, Shijingshan), which occupy less than 10% of the city’s total land area but accommodate more than 50% of its population. This region is characterized by high density and intense daily travel demand. Composite walkability scores were discretized into five performance classes from I (0.801−1.000) to V (0.000−<0.200), corresponding to “very high”, “high”, “medium”, “low”, and “very low” walkability for communication and mapping.

Results The mean composite score across the study area is 0.558, indicating moderate overall walkability with marked spatial heterogeneity. Class proportions are: I 23.69% (12,707 units), II 19.04% (10,210), III 26.46% (14,190), IV 16.63% (8,921), and V 14.17% (7,603). A clear ring-pattern emerges: scores are higher between the Second and Fourth Ring Roads, but lower within the inner core and at peripheral edges. In the historic inner city—despite dense networks and strong transit—comfort sub-scores are depressed by constrained built forms and heritage alleys, while functions concentrate in government or cultural uses. Peripheral zones perform relatively better on greenery and basic safety yet lag in land-use diversity, accessibility, and street activity. High-performing units cluster as patches that become more continuous toward the center, prominently around commercial and mixed-service hubs (e.g., CBD and Guomao, Financial Street, Zhongguancun, Sanlitun, Olympic area). Major ring roads and radial expressways exhibit low walkability, interrupting otherwise contiguous high-score corridors.Accuracy of assessment results was tested on three representative streets—Xueyuan Road, Wangfujing Street, and Sanlitun Road—using a random street-intercept survey of 1,217 pedestrians. Ratings used a 10-point scale and were normalized to 0,1. Inter-rater reliability was strong, intraclass correlation coefficient (ICC) = 0.817, 95% confidence interval (CI) = 0.79−0.85, p < 0.001. Model-to-perception fit was high: mean absolute error (MAE) was 0.0734 (Xueyuan Road), 0.0811 (Wangfujing Street), and 0.0839 (Sanlitun Road); mean absolute relative error was lowest on Sanlitun (0.112); and prediction variance—standard deviation (SD) of errors, was smallest on Wangfujing (0.092). Collectively, the low error magnitudes demonstrate that the computational scores closely track lived pedestrian experience across distinct street typologies.Based on the assessment results, several major challenges in the walkability of Beijing’s central urban districts can be identified. 1) Low walkability in peripheral areas: Streets located in the outer edges of the central districts generally perform poorly, particularly in the northeast, southwest, and northwest sectors. 2) Scattered distribution of very high walkability streets: Streets classified as “very high” in walkability are dispersed in isolated patches across the central area, lacking continuity and reducing their capacity to guide surrounding improvements. 3) Low and uneven vitality: Vitality scores are generally low and display strong spatial polarization, with the urban core performing much better than the outer areas. 4) Imageability and comfort require further improvement: While moderate overall, these two dimensions exhibit a spatial imbalance, being significantly stronger in the urban core than at the periphery.

Conclusion By integrating multi-source urban datasets, deep learning techniques (DeepLab v3+, YOLO v5), GIS analytics, and analytic hierarchy process-based (AHP-based) weighting, this research develops a comprehensive and detailed framework for assessing street walkability, rigorously validated against pedestrian perceptions. The method efficiently evaluates large street networks, captures diverse dimensions—from attraction and comfort to perceived safety—and offers quantitative support for street renewal and pedestrian environment enhancement, as well as scientific evidence for policy-making and urban spatial design.