Review On Image based Coffee Bean Quality Classification: Machine Learning Approach

Specialty coffee’s demand is growing worldwide as coffee drinkers continue to look for the freshest and highest-quality flavors. Depending upon the quality, there are two categories in the coffee industry, that is specialty coffee and commodity/commercial coffee. Coffee beans are graded via visual inspection and cupping. A 300g sample of green coffee beans is used for visual assessment, and faulty beans are counted. As per the ‘Specialty Coffee Association of America’ (SCAA), defect can be either primary or secondary. For a coffee to be a specialty, it should have less than 5 secondary defects and zero primary defects. In this survey we have presented the coffee bean quality-related research which includes various machine learning approaches in classifying the coffee beans. The study has achieved quite promising prediction accuracies and was evaluated with test data. We have done a study on coffee bean quality classification and are willing to contribute an arabica coffee bean dataset and detection of coffee bean quality using transfer learning with higher accuracy.