Mapping the Nutritional Foodscape using Crowdsourced Food Images: Case of Hartford

Keywords: community food environment, deep learning, image recognition, volunteered geographic information, GeoAI
Session Type: Virtual Paper Abstract
Day: Friday
Session Start / End Time: 2/25/2022 03:40 PM (Eastern Time (US & Canada)) - 2/25/2022 05:00 PM (Eastern Time (US & Canada))
Room: Virtual 44

Abstract

Existing research on the community food environment emphasizes the spatial dimension of food access, where food quality and nutrition are overlooked. This void can be filled by advances in deep learning models through food image recognition. Deep learning models can recognize the food item in an image and derive its associated nutrition information, including calories, macronutrients (carbohydrates, fats, and proteins), and micronutrients (vitamins and minerals). This technology has yet to be implemented for the assessment of the community food environment. In this paper, we crowdsource 15,908 food images of 470 restaurants in the Greater Hartford region on Tripadvisor and Google Place. These food images are processed by a proprietary deep learning model for food nutrition assessment. The derived nutrition information is visualized at both the restaurant level and the census tract level to perform large-scale nutrition assessments in the community food environment. As a result, the study can shed insights into examining the community food environment from a nutritional science perspective and can help elucidate the linkage between food nutrition and community health.