Leverage Infrared Spectra and Machine Learning to Develop Fast Coffee Variety Classification

Coffee has become one of the daily beverages to people around the world. However, there is a significant difference in the caffeine content between different coffee varieties, such as Arabica and Robusta beans, leading to substantial differences in prices. Consequently, there have been incidents of unscrupulous traders mixing low-cost beans with high-cost ones (or counterfeit beans) for sale, significantly impacting consumer rights and health. While there are some chemical and optical inspection techniques available in the market (such as gas chromatography and liquid chromatography for testing), these methods have several drawbacks that make them less accessible to the general public: sample destructiveness, time-consuming processes, high costs, and the need for professional operators. This study utilizes the FieldSpec 4 spectrometer combined with machine learning (ML) techniques to explore the feasibility of developing rapid screening technology for coffee varieties (coffee powder, hot brewed coffee, cold brewed coffee). Preliminary results indicate that the developed technology can identify the types of coffee beans used. To the best of our knowledge, it is the first work that leverage infrared spectra and ML to develop fast coffee variety classification. The findings of this research could pave the way for further development of low-cost rapid screening technology for coffee varieties, holding significant potential in safeguarding consumer health and rights.