Coffee is one of the most popular beverages in many places. The development of coffee cultivation in Indonesia has grown rapidly, in 2016, about 633 thousand tons of coffee were produced and in 2017, 636.7 thousand tons, which means that the growth was about 0.74%. Arabica and Robusta are the coffee species that account for most of the world's coffee trade. There are differences between Arabica and Robusta coffees both in terms of taste and the shape of the beans. Many coffee sellers and vendors do not know the difference. Therefore, to avoid mistakes when choosing coffee beans, you need software that helps distinguish types of coffee beans based on their shape. As an alternative software, image processing using the artificial neural network method can be used. This study looks at how the software detects the pattern of the coffee bean image using a morphological method using quantitative learning methods. The feature vectors used were width, height, width, mean, variance and standard deviation. The class classification is divided into the image class of Arabica coffee beans and the image class of Robusta coffee beans, with an average recognition result during the testing process of 97%
