基于语义分割模型的城市街景空间评价与资本化效应研究OA

The pursuit of residential environmental quality is a concrete manifestation of residents'aspirations for a better life in the new era,and the surrounding streetscape environment,residential properties gradually become a critical consideration in home purchasing decisions.Current studies on the capitalization effects of neighborhood environments primarily focus on the configuration of public service facilities.To further explore the impact of visual streetscape environments on housing prices,this research employs streetscape data and the SAM(segment anything model)based on deep learning technology to extract visual features of streetscape around residential areas.A hedonic price model is then applied to analyze the influence of these visual features on housing prices across multiple dimensions.The findings reveal that:(1)Streetscape visual features exert a significant impact on housing prices,with variations in willingness to pay for urban streetscape among homebuyers of different purchasing power levels and different urban regions;(2)The effects of streetscape visual indicators on housing prices are nonlinear.Except for sky openness,other streetscape features exhibit threshold effects,homebuyers demonstrate the strongest willingness to pay at some specific thresholds;(3)Interactive capitalization effects exist between streetscape visual indicators and traditional housing characteristics,with streetscape visual indicators serving a complementary role to conventional features.