Ear Biometry: Protection Safeguarding Ear Acknowledgment Framework utilizing Transfer Learning in Industry 4.0

Abstract

Human-Recognition using biometric features proposes a noble way of uniquely identifying individuals considering they do not require people to keep loads of passwords in mind to prove their identity. Biometrics have aided in people unable to render other’s identity as well and have advanced over the years. Identification using ear biometric technique is considered to outrun other features since passive human involvement and ease-of-access are its strong set of attributes, not seen in any other biometric techniques. It has managed to identify criminals in the Crime Branch and has various helpful applications in the industry. Even in times of  pandemic, it can serve its purpose in identifying individuals due to its feasibility. Models for Ear Image Identification have been proposed by various researchers over time utilizing Deep Learning and its Networks achieving high accuracy results, presenting faster and accurate identification models, boosting Ear Biometry as a secure Human Identification Tool. The learnings of this study on the AMI Ear Dataset and the OCEar dataset prove Ear Uniqueness of individuals demonstrating the identification of individuals, introducing passive identification into play, as well as computes on another dataset OCEar to study the similarity of both ears for identifying an individual.