EdgeEar: Efficient and Accurate Ear Recognition for Edge Devices

Evaluation on the Unconstrained Ear Recognition Challenge (UERC2023) benchmark shows that EdgeEar achieves the lowest EER while significantly reducing computational costs. These findings demonstrate the feasibility of efficient and accurate ear recognition, which we believe will contribute to the wider adoption of ear biometrics. I. INTRODUCTION Ear recognition has gained attention as a reliable biometric modality for unobtrusive and contactless authentication [11]. Despite advances in deep learning yielding significant performance improvements, state-of-the-art models often rely on computationally intensive architectures, making them unsuitable for resource-constrained edge devices and limiting its wider use on general hardware. Addressing this gap is crucial for secure and efficient authentication across mobile, IoT, and various real-world applications. Building on the success of lightweight face recognition models such as EdgeFace [13], we introduce EdgeEar, a model Figure 1: Comparison plot of Equal Error Rate (EER) versus designed for ear recognition on edge devices.