A fast and nondestructive method for identifying liquor from outside the bottle

The development of rapid and nondestructive techniques to protect the quality and safety of liquor is essential for public health. Due to its fingerprint spectroscopy, fast acquisition and noninvasive features, Raman spectroscopy has shown great potential in food safety. Herein, we propose a normal Raman spectroscopy technique combined with convolutional neural networks (CNN) to identify liquor from outside the bottle. The results indicate that the detection technique could accurately identify different liquors from outside the bottle. In addition, this technique still provides accurate identification when the difference in flavor factor content is as low as 0.25% of the liquor mass fraction. The proposed strategy is also applicable in the identification of different brands and vintages of liquor from other countries. Therefore, normal Raman spectroscopy combined with CNN provides a new potential research approach for liquor quality.