Dark kitchens and streetscapes: Exploring the location choices of ‘dark kitchens’ using street view imagery

Abstract

The rise of online food delivery platforms has led to the emergence of "dark kitchens", which operate without dine-in spaces. Several studies have examined the locational characteristics of restaurants, but the distinctive spatial patterns of dark kitchens compared to dine-in establishments remain unclear. To address this gap, we investigated the relationships between streetscape quality and the locational patterns of dark kitchens and dine-in restaurants, with a particular focus on identifying streetscape elements as potential differentiating factors. Streetscape quality, estimated through street view imagery, semantic segmentation, and perceptual deep-learning models and reduced into key components via principal component analysis, was used in negative binomial regression models to clarify how environmental features and socioeconomic attributes affect the distribution of dark kitchens and dine-in restaurants. Our findings indicate increased locational likelihoods of dark kitchens in areas with higher comfort, characterized by enclosure and sidewalk coverage, but decreased likelihoods in areas with high aesthetic quality. Notably, dine-in restaurants exhibit strong positive associations with both vibrancy and comfort, demonstrating their dependence on active and pedestrian-friendly streetscapes, whereas dark kitchens expand in efficiency-oriented environments by strategically prioritizing logistical accessibility over visual attractiveness. This study advances location theory and provides planning insights to adapt urban policies to emerging business models, responding to the ways digital platforms are reshaping the geography of food services. In practice, the study emphasizes the need for differentiated spatial strategies that balance delivery-oriented efficiency with preserving vibrant, pedestrian-centered retail environments.