Transfer Learning: A way for Ear Biometric Recognition
Swapnil Singh, Snehil Suman · 2022 IEEE 7th International conference for Convergence in Technology (I2CT) · 2022
Biometric recognition is a way of identifying an individual based on their biological and physiological characteristics; ear lobes are one way for doing so. Considering the current pandemic, contactless recognition became the need of the hour, for which ear biometric recognition paves the way ahead. Machine learning and deep learning can classify subjects using images, showing the direction towards building ear recognition systems. This paper proposed using transfer learning on the augmented IIT Delhi dataset for classifying segmented images of subjects. Transfer learning provides the opportunity to reduce the building time of a model and gives better performance. To verify this theory, we compared the performance of Convolutional Neural Network, VGG16, and ResNet50. After the experimental study, we extrapolated that VGG16 outperformed Convolutional Neural Network and ResNet50 by giving an accuracy of 89.73%.