Research Article
Bottleneck Feature Extraction-Based Deep Neural Network Model for Facial Emotion Recognition
@INPROCEEDINGS{10.1007/978-3-030-64002-6_3, author={Tian Ma and Kavuma Benon and Bamweyana Arnold and Keping Yu and Yan Yang and Qiaozhi Hua and Zheng Wen and Anup Kumar Paul}, title={Bottleneck Feature Extraction-Based Deep Neural Network Model for Facial Emotion Recognition}, proceedings={Mobile Networks and Management. 10th EAI International Conference, MONAMI 2020, Chiba, Japan, November 10--12, 2020, Proceedings}, proceedings_a={MONAMI}, year={2020}, month={12}, keywords={Emotion recognition Deep neural network K-nearest neighbor Haar features}, doi={10.1007/978-3-030-64002-6_3} }- Tian Ma
Kavuma Benon
Bamweyana Arnold
Keping Yu
Yan Yang
Qiaozhi Hua
Zheng Wen
Anup Kumar Paul
Year: 2020
Bottleneck Feature Extraction-Based Deep Neural Network Model for Facial Emotion Recognition
MONAMI
Springer
DOI: 10.1007/978-3-030-64002-6_3
Abstract
Deep learning is one of the most effective and efficient methods for facial emotion recognition, but it still encounters stability and infinite feasibility problems for faces of different races. To address this issue, we proposed a novel bottleneck feature extraction (BFE) method based on the deep neural network (DNN) model for facial emotion recognition. First, we used the Haar cascade classifier with a randomly generated mask to extract the face and remove the background from the image. Second, we removed the last output layer of the VGG16 transfer learning model, which was applied only for bottleneck feature extraction. Third, we designed a DNN model with five dense layers for feature training and used the famous Cohn-Kanade dataset for model training. Finally, we compared the proposed model with the K-nearest neighbor and logistic regression models on the same dataset. The experimental results showed that our model was more stable and could achieve a higher accuracy and F-measure, up to 98.59%, than other methods.
