CNN-Based Facial Emotion Recognition: Modeling and Evaluation

Juntao Chen

2024

Abstract

Human facial emotion recognition is pivotal across various domains, including human-computer interaction, psychological health assessment, and social signal processing. Despite its significance, the accuracy and robustness of facial expression recognition still faces significant challenges caused by the heterogeneity of faces and changes in the imaging environment such as posture and lighting. This study introduces an effective facial expression recognition model using convolutional neural networks (CNNs). In this experiment, a straightforward CNN model was developed and trained on the CK+48 dataset, which underwent data preprocessing steps such as image scaling, normalisation, and one-hot encoding of labels. The model architecture incorporates classical CNN components including convolutional, pooling, and fully connected layers, coupled with appropriate loss functions, optimizers, and evaluation metrics. Experimental findings showcase the CNN model's remarkable performance, achieving an average accuracy of 83.24% on the emotion recognition task, with the highest recognition rate of 98.9% for the anger expression. These results underscore the vast potential of the proposed CNN-based approach in advancing facial emotion analysis and recognition applications.

Paper Citation

in Harvard Style

Chen J. (2024). CNN-Based Facial Emotion Recognition: Modeling and Evaluation. In Proceedings of the 1st International Conference on Engineering Management, Information Technology and Intelligence - Volume 1: EMITI; ISBN 978-989-758-713-9, SciTePress, pages 250-256. DOI: 10.5220/0012925500004508

in Bibtex Style

@conference{emiti24,
author={Juntao Chen},
title={CNN-Based Facial Emotion Recognition: Modeling and Evaluation},
booktitle={Proceedings of the 1st International Conference on Engineering Management, Information Technology and Intelligence - Volume 1: EMITI},
year={2024},
pages={250-256},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0012925500004508},
isbn={978-989-758-713-9},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 1st International Conference on Engineering Management, Information Technology and Intelligence - Volume 1: EMITI
TI - CNN-Based Facial Emotion Recognition: Modeling and Evaluation
SN - 978-989-758-713-9
AU - Chen J.
PY - 2024
SP - 250
EP - 256
DO - 10.5220/0012925500004508
PB - SciTePress