Comprehensive Survey of Machine Learning Techniques for Ear Recognition System
Ulka P. Patil, D. N. Besekar, Suhas Gajanan Sapate · International Journal of Advanced Networking and Applications · 2024
Ear biometrics is non contacting and so it can be applied for identification of a human at a distance, making it a helpful supplement to facial recognition, law enforcement, crime investigation etc.Although ear detection and identification systems have rapidly improved to a certain extent, their success is still confined to specific circumstances such as an occlusion of hair.A major challenge for researchers nowadays is to recognize human based on ear with pose variations and occlusion.This summarized survey aims at identifying the research gap which is helpful in proposing a novel machine learning approach as a pathway for budding researchers.Most of the selected articles have common and a wide variety of preprocessing, feature extraction techniques such as SIFT, Gabor filter, shape features are gain.Performance of all surveyed methods is evaluated for comparison purposes using evaluation metrics such as Precision, Recall and Accuracy.The challenges before an effective Ear recognition system are discuss.This comprehensive survey article will be useful for identifying research gaps as a pathway for the same researchers to get the idea regarding of Ear recognition system which further can be transformed into a marketable product.This article at the end presents the prototype model.