Smartphone platform based on gelatin methacryloyl(GelMA)combined with deep learning models for real-time monitoring of food freshness

Abstract

Real-time monitoring of food freshness remains a challenge both for food industry and consumers since no detection devices with portability, affordability and efficiency has been commercialized to date. Here, we developed a facile sensing platform based on a smartphone application (APP) with incorporation of a deep-learning model for the real-time monitoring the food freshness. The colorimetric indicator bars on a cellulose paper were firstly constructed through the gelatinization of synthesized gelatin methacryloyl (GleMA) via UV-induced crosslinking with encapsulation of bromocresol green (BCG). After taking photos, the deep-learning model with convolutional neural network (CNN) was trained using 1735 images of labeled bars and then well predicts the meat freshness with an overall accuracy of 96.2%. Meanwhile, integrating VGG 16 architecture for the CNN and marked-based watershed algorithm into a smartphone APP could make consumers recognize the meat freshness within 30 s by simply scanning the packaging. Our sensing platform was verified as sensitive, automatic and non-destructive, which has a potential application both for food industry and consumers to real-time monitor the food freshness.