A Comparison Study of Face, Gait and Speech Features for Age Estimation
With the growing importance of age estimation in the recent years, Researchers have been trying to use different human body biometrics to estimate the age of a person. Face, gait and speech are the three main biometric traits which have been reported to i
- PDF / 175,735 Bytes
- 7 Pages / 439.37 x 666.142 pts Page_size
- 50 Downloads / 192 Views
Abstract With the growing importance of age estimation in the recent years, Researchers have been trying to use different human body biometrics to estimate the age of a person. Face, gait and speech are the three main biometric traits which have been reported to investigate the human age successfully. Each feature has specific characteristics which employ the prediction of age. Like Wrinkles, skin and shape of the face; speed, head to body ratio and height of the gait; and pitch and heaviness of the speech define the baselines for the age estimation. We have compared these three features and evaluated their performance. Conventional techniques have been used from the literature and experimental results are compared in terms of MAE and accuracy. Face is found to have most detailed features to predict the age and hence minimum mean absolute error of 5.36. It is followed by gait and then speech which are found to have mean absolute error of 6.57 and 6.62 respectively. Keywords Human age estimation Comparison
Biometrics Face Gait Speech
1 Introduction Soft Biometrics [1] has emerged as a new topic in the field of biometrics which includes the estimation of demographic characteristics from the biometric organs. Soft biometrics is defined as the estimation of age, gender, ethnicity, eye-, hair-, P. Punyani (&) A. Kumar (&) Indra Gandhi Delhi Technical University for Women, Kashmiri Gate, Delhi, India e-mail: [email protected] A. Kumar e-mail: [email protected] R. Gupta (&) Ambedkar Institute of Advanced Communications and Research, Shastri Park, Delhi, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2018 A. Kalam et al. (eds.), Advances in Electronics, Communication and Computing, Lecture Notes in Electrical Engineering 443, https://doi.org/10.1007/978-981-10-4765-7_34
325
326
P. Punyani et al.
skin colour and other biological details from body parts like face, gait, fingerprints, palm prints. It may also be estimated from acoustic waves like speech patterns. Evolution of this field comes from the growing online market. Both customers and sellers demand the need of a personalized shopping experience where product choices are shown to the customers according to their age group interests and gender category. Moreover, we can use the technology of age prediction at entrance doors in order to prevent the below 18 children to watch adult movies or from buying alcohol bottles. Age and gender estimation is an alternative way of performing security identification and verification tasks when just biometric recognition does not solve the purpose. Till date, work has been done to estimate the age and categorize the gender based on face, gait and speech only. In this paper, we compare these three techniques of age estimation and report the results. Widely used state of the art techniques are used for the implementation of each technique and comparisons are made in terms of mean absolute error (MAE). Face has various features which modify with age. Like size of face is different for a chi
Data Loading...
