Authenticity assessment of goat milk: detecting dilution with cow milk by ML-enhanced speckle pattern imaging

Dilution of precious goat milk with cheaper cow milk represents one of the most common types of food fraud in the dairy industry. Existing methods for detecting the presence of cow milk mainly rely on chemical tests or spectroscopic analyses, which require dedicated laboratory facilities, expensive instrumentation, complex sample preparation, and long waiting times for results. We propose an innovative method based on the combination of speckle pattern imaging and artificial intelligence models to detect the adulteration of goat milk with cow milk. The instrumental setup we developed is based on a low-cost semiconductor laser and an industrial CMOS camera. We analyzed five milk samples and applied well-established machine learning models, achieving an accuracy of 96.9% on the test set. These results are very promising and pave the way for the development of new opto-electronic systems to fight milk adulteration.