街道建成环境中的城市非正规性 基于北京老城街景图片的人工打分与机器学习相结合的识别探索

Understanding Urban Informality in Street Built Environment Combining Manual Evaluation with Machine Learning in Processing the Beijing Old City's Street-view Images

基于街景图片数据,通过将人工打分与机器学习结合,文章试图建立城市非正规性这一非物质环境要素在街道中的空间表征识别与评价方法体系。在北京老城街道的实证分析中,对于街道上哪些建成环境要素更能准确反映城市非正规性,基于人工打分的图像识别结果比机器学习的结果更加准确。城市非正规性在北京老城街道中的空间表征间接反映出北京老城非正规性就业群体和居民日常生活行为发生地点的微观空间分布特征。经过街景图片与实际地点的比对,三种类型的街道不具有城市非正规性特征:一是两侧已经过老旧平房改造,现状为现代居住小区或单位用房的街道;二是历史上达官显贵居住的地区,现留存有较多文保单位;三是环境较为破败的胡同,其中没有商贩经营,也没有居民自发在街道上形成的休闲空间。这一结论可作为后续城市非正规性和城市贫困关联研究的基础。

By combining the models of manual evaluation and machine learning, in the case of Beijing old city＇s street-view images, this paper tries to formulate a methodology that identifies and analyzes urban informality. The paper has two main findings. The first is that manual evaluation is more accurate than machine learning in terms of physical features that manifest urban informality in street built environment. The second is that spatial representation of urban informality in Beijing old city indicates the microscopic spatial distribution of informally employed groups and local residents＇ daily activities. By comparing street-view images and corresponding locations, it is identified that three types of street built environments have no sign of urban informality： former historical districts that have become new urban blocks through regeneration process; areas once inhabited by the privileged class in the past and left with large amount of historic heritages; run- down laneways with no activity of street vendors and recreational space for local residents. The conclusion of the study lays the foundation for future research of the relationship between urban informality and urban poverty.