TY - GEN
T1 - Facial Expression Recognition with Machine Learning
AU - Chang, Jia Xiu
AU - Poo Lee, Chin
AU - Lim, Kian Ming
AU - Yan Lim, Jit
N1 - Publisher Copyright:
© 2023 IEEE.
PY - 2023
Y1 - 2023
N2 - Human facial expressions play a crucial role in communication and enhancing interactions between humans and computers. This paper presents a novel approach for facial expression recognition using an ensemble classifier consisting of pre-trained models and vision transformers. The ensemble classifier comprises four models: VGG-19, VGGFace, ViT-B/16, and ViT-B/32. To evaluate the performance, the ensemble classifiers employ hard majority voting on three widely-used public datasets: CK+, FER2013, and JAFFE. The experimental results demonstrate that our proposed ensemble classifiers surpass the state-of-the-art methods across all datasets. Notably, we achieve outstanding accuracy rates, reaching 100% accuracy on the cleaned CK+ dataset, 76.30% accuracy on the cleaned FER-2013 dataset, and 100% accuracy on the cleaned JAFFE dataset.
AB - Human facial expressions play a crucial role in communication and enhancing interactions between humans and computers. This paper presents a novel approach for facial expression recognition using an ensemble classifier consisting of pre-trained models and vision transformers. The ensemble classifier comprises four models: VGG-19, VGGFace, ViT-B/16, and ViT-B/32. To evaluate the performance, the ensemble classifiers employ hard majority voting on three widely-used public datasets: CK+, FER2013, and JAFFE. The experimental results demonstrate that our proposed ensemble classifiers surpass the state-of-the-art methods across all datasets. Notably, we achieve outstanding accuracy rates, reaching 100% accuracy on the cleaned CK+ dataset, 76.30% accuracy on the cleaned FER-2013 dataset, and 100% accuracy on the cleaned JAFFE dataset.
KW - Convolutional Neural Network (CNN)
KW - Ensemble models
KW - Facial expression recognition
KW - Vision transformers
UR - http://www.scopus.com/inward/record.url?scp=85174384289&partnerID=8YFLogxK
U2 - 10.1109/ICoICT58202.2023.10262748
DO - 10.1109/ICoICT58202.2023.10262748
M3 - Conference contribution
AN - SCOPUS:85174384289
T3 - International Conference on ICT Convergence
SP - 125
EP - 130
BT - 2023 11th International Conference on Information and Communication Technology, ICoICT 2023
PB - IEEE Computer Society
T2 - 11th International Conference on Information and Communication Technology, ICoICT 2023
Y2 - 23 August 2023 through 24 August 2023
ER -