Using Google Street View imagery, we explore the capabilities of a multimodal large language model and semantic segmentation to construct neighborhood measures that accurately: (i) identify the ...