Abstract
Coffee industries need to maintain their product's flavour and characteristics in order to retain the product's brand and image in the market. Flavour and coffee quality during brewing is closely related to its unique roasting degree. Thus, a stable, standardized, coffee roasting degree throughout the production stages are required. In this paper, we propose a quality assessment system driven by deep learning to help monitoring the roasting process remotely and accurately and to predict the roasting degree of coffee bean A proposed deep learning-based solution for coffee roasting quality assessment is embedded into android smartphones for on-demand portable usage and support for classification insight on roasting degree The experiment initially prepared coffee roasting process dataset consisted of three classes. The results reveal that MobileNetV2 is best suited to use for roasting quality classification. Best inference time ranged between 44-50 ms using CPU and 34-44 using GPU using selected android smartphones It indicated that real-time monitoring during on-demand quality assessment in MobileNetV2 model achieved 97,75% accuracy, 96,44% recall, and 96,33% Precision on average.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have