A helmet is a protective device that is worn on the head and made of metal or other hard materials. Indonesia is a country that requires motorcyclists to wear helmets. Lack of public awareness in driving using a helmet can endanger themselves and others. For this reason, an information technology system is necessary to monitor traffic activities 24 hours a day. This study designed an application to detect the use of helmets and classify motorists using helmets or not. The method used in detecting objects in the head area uses the Convolutional Neural Network (CNN) method with the You Look Only Once (YOLO) Algorithm. This system is able to detect violations committed by motorcycle riders. The accuracy of detecting helmet use with the training dataset as a test method of model evaluation produces predictions with an average accuracy rate of 89.04%, and the avg_loss training yields a rate of 1.2%.