Maximum Entropy Binary Encoding for Face Template Protection

TL;DR

This paper introduces a novel face template protection method using deep neural networks to generate maximum entropy binary codes, which are then hashed to ensure security, cancelability, and high matching accuracy in face authentication.

Contribution

It presents a new framework combining deep CNNs and maximum entropy binary codes for secure face template protection with state-of-the-art performance.

Findings

Achieved ~95% genuine accept rate at zero false accept rate.

Provided up to 1024 bits of template security.

Validated on multiple face databases with high accuracy.

Abstract

In this paper we present a framework for secure identification using deep neural networks, and apply it to the task of template protection for face authentication. We use deep convolutional neural networks (CNNs) to learn a mapping from face images to maximum entropy binary (MEB) codes. The mapping is robust enough to tackle the problem of exact matching, yielding the same code for new samples of a user as the code assigned during training. These codes are then hashed using any hash function that follows the random oracle model (like SHA-512) to generate protected face templates (similar to text based password protection). The algorithm makes no unrealistic assumptions and offers high template security, cancelability, and state-of-the-art matching performance. The efficacy of the approach is shown on CMU-PIE, Extended Yale B, and Multi-PIE face databases. We achieve high (~95%) genuine accept rates (GAR) at zero false accept rate (FAR) with up to 1024 bits of template security.