Evaluating the spatial quality of a living street entails identifying and assessing the outdoor space that influences residents’ leisure and recreation, which may contribute to urban renewal. The application of multi-source data and deep learning technology enables an objective evaluation of large-scale spatial quality as opposed to the traditional questionnaire survey or experts’ subjective evaluation. Based on street view images, points of interest, and road network data, this study developed subjective and objective evaluation indicators for the central city of Hengyang using semantic segmentation and ArcGIS spatial analysis. This study then assigned weights to each indicator and calculated the spatial quality score for living streets. In addition, the subjective evaluations of the street view images were compared to test and verify the validation of the objective evaluation model. Finally, the study analyzed the accessibility within 500 m of the study area using Spatial Syntax and ArcGIS to overlay the low spatial quality score with the highest accessibility to identify the streets with the highest priority in the subsequent urban plan. The results indicate that the spatial quality of living in the west of Hengyang is higher than that in its northeast region. In addition, Xiao Xia Street, Guanghui Street, and Hengqi Road comprised the majority of the areas that required a priority update. Correspondingly, our research is expected to be a useful management tool for identifying urban street space issues and guiding urban renewal.