Automated ethnicity recognition using equilibrium optimizer with machine learning on facial images

Abstract

In recent times, computer vision related face image analysis has gained significant attention in various applications namely biometrics, surveillance, security, data retrieval, informatics, etc. The main objective of the facial analysis is to extract facial soft biometrics like expression, identity, age, ethnicity, gender, etc. Of these, ethnicity recognition is considered a hot search topic, a major part of community with deep connections to many social and ecological concerns. The deep learning and machine learning methods is merit for effective ethnicity classification and recognition. This study develops a facial imaging based ethnicity recognition using equilibrium optimizer with machine learning (FIER-EOML) model. The goal of the FIER-EOML technique is to detect and classify different kinds of ethnicities on facial images. To accomplish this, the presented FIER-EOML technique applies an EfficientNet model to generate a set of feature vectors. For ethnicity recognition, the presented model uses long short-term memory method. To improve the recognition performance, the FIER-EOML technique utilizes EO algorithm for hyperparameter tuning process. The performance validation of the FIER-EOML technique is tested on BUPT-GLOBALFACE dataset and the results are examined under several measures. The comprehensive comparison study reported the enhanced performance of the FIER-EOML technique over other recent approaches.