Name
Abstract | ||
|---|---|---|
This article proposes a noncontact emotion recognition method based on complexion and heart rate (HR). Unlike most existing noncontact emotion recognition methods, our method uses complexion and heart rate as contactless emotion recognition indicators, effectively avoiding the possibility of disguised facial expressions or acoustic features in noncontact methods. Different from most existing contact emotion recognition methods, our method only employs an ordinary device, Kinect, to achieve the complexion and heart rate, which can effectively avoid subjects' psychological oppression when using wearable devices. Convolutional neural network (CNN) and bidirectional long short-term memory-conditional random fields (Bi-LSTM-CRF) are developed to extract the features of complexion and heart rate for contactless emotion recognition, respectively. We experimentally compared our methods with the existing five emotion recognition methods in practical application. The results show that our methods can outperform the existing five methods and obtain an emotion recognition accuracy higher than 80% in detection of anger, depression, doubt, and indignation. This indicates that the fusion of complexion and heart rate can effectively improve the emotional recognition accuracy. Note that our method requires neither expensive equipment nor conscious emotional acquisition by the participant, thereby promoting the development of emotion recognition in various fields such as the interrogation process and psychological counseling. |
Year | DOI | Venue |
|---|---|---|
2022 | 10.1109/TIM.2022.3194858 | IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT |
Keywords | DocType | Volume |
Emotion recognition, Heart rate, Feature extraction, Psychology, Electroencephalography, Electrocardiography, Employee welfare, Bidirectional long short-term memory-conditional random fields (Bi-LSTM-CRF), complexion, convolutional neural networks (CNNs), emotion recognition, heart rate (HR), information fusion, recurrent neural network (RNN) | Journal | 71 |
ISSN | Citations | PageRank |
0018-9456 | 0 | 0.34 |
References | Authors | |
0 | 6 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Guanglong Du | 1 | 0 | 2.37 |
| Qinglin Tan | 2 | 0 | 0.34 |
| Chunquan Li | 3 | 95 | 12.61 |
| Wang X | 4 | 25 | 14.20 |
| Shaohua Teng | 5 | 151 | 34.68 |
| Peter X. Liu | 6 | 1 | 2.04 |