Research Article
Wearable-Based Human Emotion Inference System
@INPROCEEDINGS{10.1007/978-3-031-04231-7_12, author={Zirui Zhao and Canlin Zheng}, title={Wearable-Based Human Emotion Inference System}, proceedings={Edge Computing and IoT: Systems, Management and Security. Second EAI International Conference, ICECI 2021, Virtual Event, December 22--23, 2021, Proceedings}, proceedings_a={ICECI}, year={2022}, month={5}, keywords={Wearable devices Emotion recognition Multimodal data Multiple perception}, doi={10.1007/978-3-031-04231-7_12} }- Zirui Zhao
Canlin Zheng
Year: 2022
Wearable-Based Human Emotion Inference System
ICECI
Springer
DOI: 10.1007/978-3-031-04231-7_12
Abstract
Many emotion recognition methods which have been proposed today have different shortcomings. For example, some methods use expensive and cumbersome special-purpose hardware, such as EEG/ECG helmets, while others based on cameras and speech caused risk of privacy leakage. With the prosperous development and popularization of wearable devices, people tend to be equipped with multiple smart devices, which provides potential opportunities for lightweight emotional perception. Based on this, we take actions on developing universal portable system and multi-source wearable sensing technology devices.
This paper designs an emotion recognition framework called MW-Emotion (Multi-source Wearable emotion recognition) featuring on low cost, universality, and portable commercial wearable devices to perceive multi-source sensing data and implement a system. It takes four basic emotions as the research object and implement an emotion recognition to explore deep context innovatively. The experimental results show that MW-Emotion has a recognition accuracy of 85.1% for person-dependent mode. The framework uses the method that different types of data are effectively fused through signal processing. We call it multimodal data fusion technology, which reduces the energy waste caused by data redundancy and effectively resists interference.
