| dc.description.abstract | Coffee (Coffea sp.) is the largest tropical commodity due to its beans that can be processed into beverages. In Indonesia, there are three most sought-after types of coffee beans, those are Arabica, Robusta, and Liberica. The characteristics of these three types of coffee beans are so similiar, making it challenging to distinguish them manually. The possibility of fraudulent coffee transactions is evident by mixing different types of coffee beans. Skill, experience, and precision are required to differentiate coffee bean types manually. Technological advancements can be utilized to support the differentiation process through the application of computer vision technology. In this study, the Single Shot MultiBox Detector-MobileNet (SSD-MobileNet) method is implemented to perform the task of detecting coffee bean types, consist of Arabica, Robusta, and Liberica. The detection of coffee bean types is carried out in real-time using a smartphone camera. Based on the conducted testing, it is found that the system built with the implementation of the SSD-MobileNet method is capable of detecting coffee bean types with an accuracy of 91,42%. | en_US |
