Currently, the use of artificial intelligence is growing rapidly, including being used to recognize human facial expressions. Human facial expressions have a complex recognition rate. In this study, deep learning will be applied to find out how much accuracy the recognition of facial expressions. The method used in this study is a combination of Viola Jones and Convolutional Neural Network. Viola Jones is used at the segmentation stage and Convolutional Neural Network to classify data. The facial expression dataset that was analyzed consisted of happiness, anger, disgust, sadness, fear, surprise and normal totaling 2205 data. Tests conducted using a conffusion matrix with an accuracy rate of 96.14%. The results of this test indicate that the proposed method has good accuracy for recognizing facial expressions.