Detection of tartrazine colored rice flour adulteration in turmeric from multi-spectral images on smartphone using convolutional neural network deployed on PaaS cloud

Abstract

Food adulteration occurs globally, in many facets, and affects almost all food commodities. Adulteration is not just a crucial economic problem, but it may also lead to serious health problems for consumers. Turmeric (Curcuma longa) is a world-class spice commonly contaminated with various chemicals and colors. It has also been used extensively in many Asian curries, sauces, and medications. Different traditional approaches, such as chemical and physical methods, are available for detecting adulterants in turmeric. These approaches are rather time-consuming and inaccurate methods. Therefore, it is of utmost importance to identify the adulterants in turmeric accurately and instantly. A cloud-based system was developed to detect adulteration in adulterated turmeric. The dataset consists of spectral images of turmeric with tartrazine-colored rice flour adulterant. Adulterants in weight percentages of 0 percent, 5 percent, 10 percent, and 15 percent were mixed with turmeric. A convolutional neural network (CNN) was implemented to detect adulteration, which achieved 100 percent accuracy for training and 94.35 percent accuracy for validation. The deep CNN (DCNN) models, namely, VGG16, DenseNet201, and MobileNet, were implemented to detect adulteration. The proposed CNN model outperforms DCNN models in terms of accuracy and layers. The CNN model is deployed to the platform as a service (PaaS) cloud. The deployed model link can be accessed using a smartphone. Uploading the adulterated turmeric image to a cloud link can analyze and detect adulteration.