The MUSKETEER project: milk adulteration detection using speckle pattern and machine learning

The MUSKETEER (Milk adUlteration detection using SpecKlE paTtern and machinE lEaRning) project aims to address the global challenge of fighting milk adulteration, which poses significant health risks for consumers. Traditional methods for milk analysis (eg. ELISA, PCR, chemical tests) are complex, time- and money-consuming. The project goal is the development of a user-friendly platform that employs real-time artificial intelligence (AI)based processing of speckle pattern (SP) images to identify adulteration in milk samples non-invasively. SP is the interference pattern produced when laser light illuminates a milk sample, which has a non-uniform refractive index distribution due to the presence of suspended particles. Images of SP acquired by an low-cost industrial camera are rich in information about the sample. In this work, we report an effective method to recognize different types of commercial cow milk and to identify milk dilution with water and 12.5% water-glucose solution. The average intensity and the dimension of the SP grains can be extracted from SP images. By considering both statistical parameters, our system can distinguish between different types of milk and detect diluted samples with both water and glucose, offering a reliable approach to address milk adulteration and ensure the integrity of dairy products on the market.