Elsevier

Gait & Posture

Volume 58, October 2017, Pages 428-432
Gait & Posture

Full length article
Self-esteem recognition based on gait pattern using Kinect

https://doi.org/10.1016/j.gaitpost.2017.09.001Get rights and content

Highlights

  • We used Kinect to capture real-time gait data to predict individual’s self-esteem.

  • Gait pattern can be used to predict self-esteem to some extent.

  • The gait predicting model can be taken as a good supplementary method to measure self-esteem.

Abstract

Background

Self-esteem is an important aspect of individual’s mental health. When subjects are not able to complete self-report questionnaire, behavioral assessment will be a good supplement. In this paper, we propose to use gait data collected by Kinect as an indicator to recognize self-esteem.

Methods

178 graduate students without disabilities participate in our study. Firstly, all participants complete the 10-item Rosenberg Self-Esteem Scale (RSS) to acquire self-esteem score. After completing the RRS, each participant walks for two minutes naturally on a rectangular red carpet, and the gait data are recorded using Kinect sensor. After data preprocessing, we extract a few behavioral features to train predicting model by machine learning. Based on these features, we build predicting models to recognize self-esteem.

Results

For self-esteem prediction, the best correlation coefficient between predicted score and self-report score is 0.45 (p < 0.001). We divide the participants according to gender, and for males, the correlation coefficient is 0.43 (p < 0.001), for females, it is 0.59 (p < 0.001).

Conclusion

Using gait data captured by Kinect sensor, we find that the gait pattern could be used to recognize self-esteem with a fairly good criterion validity. The gait predicting model can be taken as a good supplementary method to measure self-esteem.

Introduction

Self-esteem is considered as the baseline self-view from which fluctuations may occur in a given context [1], or refers to an individual’s subjective evaluation of his/her worth as a person [2], or is the evaluative component of self knowledge [3]. Self-esteem prospectively predicts success and well-being in life domains such as relationships, work, and health [4]. On the other hand, previous studies have consistently observed lower explicit self-esteem are associated higher levels of depression and anxiety symptoms [5], [6]. In short, self-esteem is a central focus of research related to mental health, personality and self-evaluation.

Although self-report questionnaire is the most widely used method of psychometric measurement including self-esteem measurement. In addition to self-reporting, observer-reporting and behavioral assessment are also potential methods to measure self-esteem. It can't be said that behavioral assessment is better than self-report, but the combination of multiple methods can improve the validity of the measurement [7]. Moreover, when subjects are not able to complete self-report questionnaire, such as children [8], people without cognitive ability, and illiteracy, behavioral assessment will be a good supplement.

The main barrier of exploiting non-verbal behavior clues to measure psychological characteristics is the difficulty of behavior quantification for human evaluator. While with the development of pattern recognition, it becomes possible to recognize psychological characteristics by using some behavioral indicators such as expression, gesture, posture, and gait. Gait is the way of people moving, which communicates a wealth of information about emotional state, cognition, intent, personality, and attitude [9]. Several studies indicate that gait is associated with personality [10], self-efficacy [11], and depression [12]. Roether et al. found critical features for the perception of emotion from gait [13], and personalities and health can also be identified by face and gait cues [14]. Though self-esteem is an internal, subjective phenomenon and is not explicitly defined as acts, some evidence has supported the possibility of measuring self-esteem by nonverbal behavior. In Kilianski’s study, perceivers can give judgments of target self-esteem after a ten minutes interview [15]. Korooshfard et al. found a negative medium relationship between round shoulder (RS) and self-esteem (SE) (r = 0.35) [16]. Furthermore, Tracy and Robins indicate that pride can be reliably assessed from nonverbal behaviors [17]. Researches on non-verbal behavior and confidence demonstrate that non-verbal cues expressing confidence or its lack accounted for ten times more than the variance due to the verbal cues [18]. Meanwhile, pride and confidence both have a strong relationship with self-esteem.

In this study, we want to build predicting models to recognize self-esteem based on gait pattern. To build the predicting models, behavioral features are extracted from individuals’ gait record in the first step. Second, these features are used as input in machine learning algorithms to predict the outcome of self-esteem. Finally, the predicting self-esteem will be compared with individuals’ self-report self-esteem to evaluate the validity of models. We hypothesize that it’s feasible to measure self-esteem by gait, which will be a new measure method that can be served as a complement of traditional measurement.

Microsoft Kinect is a system which is low-cost, portable, and does not require any sensors to be attached to the body [19]. The RGB image camera of Kinect allows it to capture 3D body gestures in real-time. The validity of Kinect has been demonstrated in the studies of gesture and motion recognition. Fern’ndez-Baena et al. found it perform well in tracking simple stepping movements [20], and Auvinet et al. successfully detected gait cycles in treadmill by Kinect [21]. Weber et al. reported that Kinect is capable of describing walking and running with acceptable accuracy in the frontal plane [22]. And Kinect was also taken as a tool in measuring clinically relevant movements in people with Parkinson’s disease [23]. These studies mentioned above motivate us to use Kinect to capture the gait.

Section snippets

Subjects

182 graduate students from University of Chinese Academy of Sciences participate this study, each participant is paid 200 RMB as compensation. The individuals are excluded if they report any injury or disability affecting the walking, and in fact no one is excluded in this experiment. After excluding the void data, there are 178 valid cases enter the data analysis, including 100 males and 78 females. All these students are in Grade one, and the average age is 24.2 (SD = 1.5).

Materials and instruments

Rosenberg Self-Esteem

Data collection

The average score on RSES is 30.9 (SD = 3.8), and for each participant we collect 3600 frames gait data for two minutes.

Data preprocessing

Firstly, we run Gauss Filter which is a kind of low-pass filter [26] to remove noises and burrs contained in the original recordings of gaits. We apply Gauss Filter to each axis (X, Y, Z) of the 25 joints. The length of the window is 5 and the convolution kernel of Gauss Filter c = [1,4,6,4,1]/16. The convert process is defined as:Out[i]=116(In[i]×1+In[i+1]×4+In[i+2]×6+In[i+3]×4+In

Discussion

This study uses the gait and self-esteem data of the subjects to train the machine learning model, which can automatically use gait data to predict self-esteem. Since we using the predict model as a new self-esteem measurement method, the correlation coefficient can be explained as criterion validity. The validity of this new method is 0.45, which can be achieved by models epsilon-SVR and nu-SVR. The result indicates that gait pattern can be used to predict self-esteem fairly well. Consistent

Conclusion

By using Kinect sensor devices, we could capture real-time walking behavioral data, and predict individual’s scores on self-esteem. The result indicates that gait pattern can be used to recognize self-esteem with a fairly good criterion validity. The predicting model can be taken as a good supplementary method to measure self-esteem.

Conflicts of interest

None.

Acknowledgements

The authors gratefully acknowledges the generous support from National Basic Research Program of China (2014CB744600), Key Research Program of Chinese Academy of Sciences (CAS) (KJZD-EWL04), and CAS Strategic Priority Research Program (XDA06030800).

References (29)

  • L.A.V. Tuijl et al.

    Implicit and explicit self-esteem and their reciprocal relationship with symptoms of depression and social anxiety: a longitudinal study in adolescents

    J. Behav. Ther. Exp. Psychiatry

    (2014)
  • J.F. Sowislo et al.

    Does low self-esteem predict depression and anxiety? A meta-analysis of longitudinal studies

    Psychol. Bull.

    (2013)
  • L. Schneider et al.

    Self-informant agreement in well-being ratings: a meta-analysis

    Soc. Indic. Res.

    (2009)
  • Z. Wang et al.

    Observed personality in childhood: psychometric and behavioural genetic evidence of two broad personality factors

    Eur. J. Pers.

    (2013)
  • Cited by (37)

    • Psychometric evaluation of the Rosenberg Self-Esteem Scale in primary school students with mild intellectual disability: First evidence

      2021, Research in Developmental Disabilities
      Citation Excerpt :

      Self-esteem is a valuable personal asset because it has been positively associated with, among other aspects, psychological well-being and happiness (Baumeister, Campbell, Krueger, & Vohs, 2003). Conversely, low self-esteem has been associated with various factors that affect a person’s mental health (Sun, Zhang, Liu, Hu, & Zhu, 2017). A child’s self-esteem is considered to be their subjective assessment of their value as a person, and is essentially determined by the amount of worth children assign to themselves (Baumeister et al., 2003).

    View all citing articles on Scopus
    View full text