Gender Identification Using 2D Ear Biometric

Due to the advancement of technologies, organizations, and governments increasingly rely on digital technologies to validate a person's identity for secure interactions and to provide services for digital access control. Critical transactions require accurate and reliable authentication and verification of a person to prevent fraud. The most reliable and secure digital technology is biometrics. In this work, we are analyzing ear biometrics. Previous studies in biometrics have shown that gender can be recognized by face and voice. In this paper, ear biometrics are used to recognize and identify genders. Machine learning and deep learning are explored for gender identification. Mask RCNN and Grabcut segmentation techniques are used for ear detection, and features are extracted from Histogram-Oriented Gradient and Gabor filters and classified by gender using machine learning and deep learning techniques. The machine and deep network models achieved recognition rates of 81.48% and 87.52% for the EarVN1.0 dataset, respectively. The proposed automated system is useful to recognize people using ear biometrics.