Facial expression recognition is an important research area in the advancement of machine learning. This research uses Convolutional Neural Network (CNN) as a method for recognizing facial expressions with a fairly high level of accuracy. This research uses a dataset obtained from Kaggle in the form of images of facial expressions, including surprised, happy, sad, afraid, angry and neutral. The MobilenetV2 CNN model was trained and tested using this dataset. The research results show that the model is able to recognize facial expressions with 78% accuracy on test data. It can be concluded that the MobilenetV2 model is quite capable of recognizing facial expressions.