Facial Expression Classification Using the Convolutional Neural Network (CNN) Method with a Comparison of Two Modified Models

Abstract

Abstract

Facial expression is an important aspect in conveying feelings. In today's era it is easier to categorize human emotions by identifying facial expressions. One method for categorizing facial expressions is deep learning. Convolutional Neural Network (CNN), which functions to process image data and detect objects in the dataset, is the most important deep learning method for understanding datasets today. In this research, the dataset used is a small and good CK+ type dataset. The aim of this research is to determine and compare the accuracy of two CNN models that have been modified by researchers, namely model A and model B, to identify facial expressions using the CK+ type dataset, because one of the main challenges in using the CNN method is that the accuracy achieved must be high. Model A is more effective than model B because it has higher accuracy, namely model A is 98.98% while model B is 91.88%. Although model B has more complex and heavy layers, this shows that model A is better at recognizing and classifying facial expressions.