A Spectral Dataset of both Liquid and Solid Milk with Impurities for some Milk Impurity Detection

This dataset was developed for a study focused on the fast, cost-effective, and accurate detection of milk impurities using machine learning algorithms in conjunction with Near-Infrared (NIR) Spectroscopy. Spectral data were gathered using pixel sensors equipped with on-chip filtering technology, incorporating up to 8 wavelength-selective photodiodes within a compact 9x9mm array. The spectrometer divided the spectrum into eight distinct bands, ranging from 400 to 1100 nm, to detect pure or impure milk for various additives in both liquid and solid forms. The additives analyzed include refined sugar, flour, coffee powder, and milk powder. The dataset comprises a total of 20,000 spectral samples.