Images of ears from WUTEarDMD database.

The high performance of deep neural networks approaches in the image and the face classification task led us to propose a solution crafted to ear identification defined by the authors of Unconstrained Ear Recognition Challenge 2019. As in this particular case it was necessary to deal with the unconstrained data matching in the open-set scenario we employed the siamese network in its core. We propose a hybrid loss function that consists of a classification-learning component (based on cross-entropy loss function) and a distance-learning component (contrastive loss function). The novelty of our approach lays in the concept of training the first function with respect to object-dependent features (specific to ear) and the second part with respect to the user (features specific to person).

Detailed description in papers:

1. Žiga Emeršič, Aruna Kumar S. V., B. S. Harish, Weronika Gutfeter, Jalil Nourmohammadi Khiarak, Andrzej Pacut, Earnest Hansley, Mauricio Pamplona Segundo, Sudeep Sarkar, Hyeonjung Park, Gi Pyo Nam, Ig-Jae Kim, Sagar G. Sangodkar, Ümit Kaçar, Murvet Kirci, Li Yuan, Jishou Yuan, Haonan Zhao, Fei Lu, Junying Mao, Xiaoshuang Zhang, Dogucan Yaman, Fevziye Irem Eyiokur, Kadir Bulut Özler, Hazım Kemal Ekenel, Debbrota Paul Chowdhury, Sambit Bakshi, Pankaj K. Sa, Banshidhar Majhi, Peter Peer, Vitomir Štruc The Unconstrained Ear Recognition Challenge 2019, CoRR, abs/1903.04143, 2019
2. J. Nourmohammadi-Khiarak and A. Pacut. An ear anti-spoofing database with various attacks, In 2018 International
   Carnahan Conference on Security Technology (ICCST), pages 1–5, Oct 2018.