This research aims to develop a digital image analysis system that can determine the level of maturity of roasted coffee beans, in terms of detecting roasted coffee beans that are suitable and not suitable for consumption and sold as quality coffee (special coffee) as stated in the coffee bean classification standards provided by SNI No. 01-2907-1999. This research aims to develop a digital image analysis system that can determine the level of maturity of roasted coffee beans in terms of color. The color feature extraction used in this research is the HSV (Hue Saturation Value) color space. This research began by collecting data in the form of 2D digital images of roasted coffee beans. The system developed in this research consists of two main stages, namely training and testing. The amount of coffee bean image data used was 90 images. The data used is in the form of images of coffee beans consisting of three levels, namely dark, light and medium. Classification uses the Naive Bayes algorithm. Based on the results of research on the analysis of coffee bean maturity levels, the highest training accuracy was 100% and the highest testing accuracy was 100%.