SP-MILK: a dataset of speckle pattern images of pure and adulterated animal milk

This dataset contains a total of 12 000 speckle pattern images of milk samples captured using an optoelectronic setup involving a CMOS camera and a red laser source. The data is divided into two sub-datasets: the "unadulterated" samples refer to milk samples without adulteration, while the "adulterated" samples refer to milk samples obtained by diluting pure cow milk with water and a mix of water and glucose. The tested samples are the following:

Unadulterated dataset:

• R: powdered dextrose (reference sample)
• S1: whole cow milk UHT
• S2: partially skimmed cow milk UHT
• S3: skimmed cow milk UHT
• S4: whole goat milk UHT
• S5: partially skimmed goat milk UHT

Adulterated dataset:

• RAD: powdered dextrose (reference sample)
• SAD1: whole cow milk UHT
• SAD2: whole cow milk UHT adulterated with a 15% water dilution (17 ml milk, 3 ml water)
• SAD3: whole cow milk UHT adulterated with a 30% water dilution (14 ml milk, 6 ml water)
• SAD4: whole cow milk UHT adulterated with a 15% water+glucose dilution (17 ml milk, 3 ml water+glucose)
• SAD5: whole cow milk UHT adulterated with a 30% water+glucose dilution (14 ml milk, 6 ml water+glucose)

The file "filenames_and_labels" links the specific SP filename to the class label (the sample) and also contains the folder structure, helping researchers dynamically build the full path to the images for data analysis.

Please remember that measurement noise due to the instrumentation and sample internal variabilities among preparations must be corrected using the data of the reference samples.

Additional details about the optoelectronic setup, the samples' preparation, the data acquisition procedure, and the dataset validation can be found in the related data article and scientific articles in the "References" section. Please cite them if you use this dataset for your research.