Emotion detection from Natural Walking
- Published
- Accepted
- Subject Areas
- Kinesiology, Psychiatry and Psychology
- Keywords
- sensor mining, emotion identificatiion, accelerometer sensor, smartphone
- Copyright
- © 2015 Cui et al.
- Licence
- This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, reproduction and adaptation in any medium and for any purpose provided that it is properly attributed. For attribution, the original author(s), title, publication source (PeerJ PrePrints) and either DOI or URL of the article must be cited.
- Cite this article
- 2015. Emotion detection from Natural Walking. PeerJ PrePrints 3:e1384v1 https://doi.org/10.7287/peerj.preprints.1384v1
Abstract
Emotion identification aims to automatically determine person's affective state, which has immense potential value in action tendency, health care, psychological detection and human-computer(robot) interaction. In this paper, we propose a novel method of identifying emotion from natural walking. After obtaining the three-axis acceleration data of wrist and ankle recorded by smartphone, we run a sliding average filter with different window size w, then cut actual data into slices. We extract 114 features including time-domain, frequency domain, power and distribution features from each slice, then use principal component analysis (PCA) for feature selection. We train SVM, Decision Tree, Multilayerperception, Random Tree and Random Forest classification models, and compare the accuracy on datasets of wrist and ankle with respect to different w. The performance of emotion identification on acceleration data of ankle is better than wrist. Among them, SVM yields the best accuracy of 90:31% for anger vs.neutral, 89:76% for happy vs.neutral, and 87:10% for anger vs.happy. The model for identifying anger/neutral/happy yields the best accuracy of 85%- 78%-78%. The results show that it is capable of identifying personal emotional states through the gait of walking.
Author Comment
Nonverbal signals are observed to deliver additional cues for a person's emotion and intention, which can be used to improve human-machine interaction or health state detection.Traditionally,emotion identification is based on facial expressions, linguistic and acoustic features in speech,which has high complexity in analyzing image and audio. Natural walking, with large amount of gaits data, offers opportunities for us to gain personnal affective state. In this paper, we propose a novel method to identify emotion from natural walking. First, we collected actual acceleration data of wrist and ankle of all subjects' natural walking. Second, we employ SVM and other machine learning algorithms to train models. Our approach achieves the best results on SVM, yielding a accuracy of 90.31% for anger vs.neural, 89.76% for happy vs.neutral, 87.10% for anger vs. happy, and 85%-78%-78% for anger/neutral/happy. Furthermore, a prototype system for identifying emotion from natural walking is deployed
