An intelligent deep learning enabled residual network for age estimation model using facial images

  • Katta Nagaraju, M.Babu Reddy

Abstract

The human face conveys important information like identity,  expression, ethnicity, gender, age, etc.Age estimation from frontal facial images becomes essential in diverse applications such as individual authentication, forensic research, security control, human–computer interaction, social media, etc. But it is difficult process owing to large differences in facial appearances with different intrinsic and extrinsic factors. Several existing techniques make use of handcrafted features for encoding the aging patterns. Recently, deep learning-based age estimation using facial images is becoming more and more important. Earlier works have focused on the usage of DL models pretrained to recognize faces, which carries out fine tuning process for precise outcomes. With this motivation, this paper presents an intelligent deep learning enabled Residual Network for age estimation model (IDLRN-AEM) using facial images. The presented IDLRN-AEM model uses the facial images as input and it employs the DL based ResNet-152 model as a featurer extractor. The presented IDLRN-AEM model carries out a feature level fusion process using the linear discriminant analysis (LDA) technique, which minimizes the dimensionalities of the feature space and determines the person’s age using kernel extreme learning machine (KELM). In order to optimally tune the ‘kernel parameter’ and ‘penalty parameter’ parameters of the KELM model, the chaotic brain storm optimization (CBSO) algorithm is employed in such a way that the classification performance can be improved.The performance validation of the IDLRN-AEM model takes place using different databases and examines the results in terms of different measures. The experimental results showed that the IDLRN-AEM model is found to be efficient over the other recent state of art methods.

Published
2021-08-12
How to Cite
M.Babu Reddy, K. N. (2021). An intelligent deep learning enabled residual network for age estimation model using facial images. Design Engineering, 8004-8022. Retrieved from http://www.thedesignengineering.com/index.php/DE/article/view/3329
Section
Articles