We used hyperspectral imaging data to characterize the consistency of 15 brands of roasted coffee beans.

Coffee brands varied markedly sampling between dates, and also within and among roasting classes.

Extractable protein content could be predicted accurately based on hyperspectral imaging data.

We provided evidence that commercial coffee brands can be classified accurately based on hyperspectral imaging data.

Results from this study are relevant to consistency and quality control of wide range of commercial food and beverage brands.