Emotion Recognition Based on Deep Learning with Autoencoder

Wiranata, I Made Nomo and Pranowo, . and Santoso, Albertus Joko (2019) Emotion Recognition Based on Deep Learning with Autoencoder. In: THE 5TH INTERNATIONAL CONFERENCE ON INDUSTRIAL, MECHANICAL, ELECTRICAL, AND CHEMICAL ENGINEERING 2019 (ICIMECE 2019). AIP Publishing, Surakarta, Indonesia, pp. 1-5. ISBN 978-0-7354-1971-1

Preview

Text (I Made Nomo Wiranata, Pranowo and Albertus Joko Santoso) 32. Emotion recognition based on deep learning with auto-encorder.pdf Download (1MB) | Preview

Abstract

Facial expression is one way of expressing emotions. Face emotion recognition is one of the important and major fields of research in the field of computer vision. Face emotion recognition is still one of the unique and challenging areas of research because it can be combined with various methods, one of which is deep learning. Deep learning is popular in the research area because it has the advantage of processing large amounts of data and automatically learning features on raw data, such as face emotion. Deep learning consists of several methods, one of which is the convolutional neural network method that will be used in this study. This study also uses the convolutional auto-encoder (CAE) method to explore the advantages that can arise compared to previous studies. CAE has advantages for image reconstruction and image de-noising, but we will explore CAE to do classification with CNN. Input data will be processed using CAE, then proceed with the classification process using CNN. Face emotion recognition model will use the Karolinska Directed Emotional Faces (KDEF) dataset of 4900 images divided into 2 groups, 80% for training and 20% for testing. The KDEF data consists of 7 emotional models with 5 angles from 70 different people. The test results showed an accuracy of 81.77%.

Actions (login required)

View Item