Personalized allergen recommender system: ingredient list scanning for safe food choices

Abstract

Understanding every ingredient in a food is crucial for everyone who has food allergies and consumes packaged foods. However, the component or ingredient lists are unable to recognized by only reading due to the lack of education related to coding conventions among the public. Despite the availability of certain allergen recommendation applications that might aid in pinpointing ingredients, the available applications are not personalized and are not supported efficiently. This research aims to develop a personalized allergen recommender system that helps with safe food selections through ingredient list scanning. The final deliverable is a mobile application that scans the list of ingredients, classifies allergen ingredients for the consumer, and provides allergen recommendations to the user. The application can identify high-risk and the most common 25 food ingredients that can cause allergic. The Tesseract OCR has been used to extract text from the captured image of ingredients. After the text is extracted, the application compares it with a user-predefined list of allergens and based on the result provides the recommendation to the user. The developed application was able to successfully extract ingredient text and accurately identify allergenic ingredients in real time. This study demonstrates that a AI -enhanced personalized allergen recommender system can improve user safety and awareness, supporting informed decisions about food products.