Deep Learning Applied on Cinnamon Powder Adulteration Detection

Cinnamon powder is an important seasoning spice frequently used in a variety of desserts. The production process of cinnamon powder is complex, which results in a relatively high price. Unscrupulous companies in the market adulterate food by using cheaper substitutes such as walnut powder or other nut powders. Apart from the problem of food fraud, the issue of allergens is even more crucial. Thus, we analyzed the adulteration of cinnamon powder through Principal Component Analysis (PCA) and using Near-Infrared Spectroscopy (NIRS). A Convolutional Neural Network (CNN) was used to detect adulteration and compare the results with those obtained by NIRS. Results indicated that the NIRS-PCA and CNN showed an accuracy rate of 99.95 and 92.80%, respectively. Finally, the proposed methods enabled quick and easy detection of adulteration in cinnamon powder, making it accessible to non-specialized personnel for testing.