Image processing and artificial intelligence approaches for food adulteration detection: The case of Antep pistachio

The safety and quality of food are of critical importance in preserving consumer health and providing a reliable product for the food industry. However, food fraud or adulteration issues worldwide have become a significant threat to the food industry. The deliberate addition of foreign substances to food or the use of low-quality ingredients can mislead consumers, leading to health risks and reliability issues. Detecting food adulteration using traditional methods can be complex and time-consuming. However, in recent years, image processing and artificial intelligence techniques have offered a faster and more effective approach to this issue. Using image processing algorithms and artificial intelligence models for detecting food fraud and controlling food quality can help manufacturers and regulators make faster and more accurate decisions. This thesis discusses the applicability of image processing and artificial intelligence techniques for detecting food adulteration in the food industry, using the example of pistachios. The proposed DenseNet model is a deep learning model with dense connections and is ideal for image-based studies. Adulterated pistachio samples were obtained, photographed, and these images were prepared for the pre-trained DenseNet model to be applied with image processing. The DenseNet model effectively classified unique datasets of pistachios and adulterated samples.