Facial expression recognition (FER) using deep learning network

Zaidi, Muhammad Irfan (2024) Facial expression recognition (FER) using deep learning network. Project Report. Universiti Teknikal Malaysia Melaka, Melaka, Malaysia. (Submitted)

Text (Full text) Facial expression recognition (FER) using deep learning network.pdf - Submitted Version Download (3MB)

Abstract

Emotion recognition plays a significant role in measuring the emotions of a person. Since our faces are the most expressive parts of our bodies and are frequently used as indicators of our mental states. This project aims to develop facial expression recognition using deep learning technique for recognize seven different emotions such as angry, disgust, fear, happy, neutral, sad and surprise. However, developing a FER system by using FER2013 dataset based on machine learning have limitation in handling complex datasets. Therefore, this project involves deep learning technique by using Convolution Neural Network (CNN) model with data augmentation to handle the complex data and be able to extract facial features from the input images with reduce overfitting. As a result, our proposed model achieved the highest accuracy with 65.27% compared to the pretrained models where VGG16 with 42.11%, AlexNet with 24.71% and MobileNet with 28.35% on the FER2013 test dataset. The FER system also can help in the healthcare and education sector. Thus, this project can be achieved the SDG goal which is SDG 3 for Good Health and Well Being and SDG 4 for Quality Education.

Actions (login required)

View Item