Human facial expressions generally represent the emotions or feelings they are feeling at that time. Classification of facial expression images can help to find out what emotions a person is feeling. CNN is type of neural network is used to extract features from an image and is very superior if implemented on image data. In this study, facial expression image classification will be carried out by applying the Convolution Neural Network to the MMA Facial Expression dataset. Where the data will be divided into 2 classes, namely happy and sad. The test is carried out using testing data for each class of the CNN model that has been created using a predetermined optimizer, namely: Adadelta, Adagrad, Adam, Adamax, Nadam, Rmsprop, and SGD. Based on the test results, it can be concluded that CNN can classify images of human facial expressions well using the SGD optimizer with an accuracy value of 63%.
Copyrights © 2021