Triplet Similarity Embedding for Face Verification

TL;DR

This paper introduces a face verification method combining deep CNNs with a low-dimensional triplet similarity embedding, achieving superior accuracy and efficiency on the challenging IJB-A dataset.

Contribution

It presents a novel triplet similarity embedding technique integrated with CNNs for face verification, improving performance and computational efficiency.

Findings

Outperforms state-of-the-art methods on IJB-A dataset

Requires less training time than existing approaches

Provides a memory-efficient embedding suitable for hashing and visualization

Abstract

In this work, we present an unconstrained face verification algorithm and evaluate it on the recently released IJB-A dataset that aims to push the boundaries of face verification methods. The proposed algorithm couples a deep CNN-based approach with a low-dimensional discriminative embedding learnt using triplet similarity constraints in a large margin fashion. Aside from yielding performance improvement, this embedding provides significant advantages in terms of memory and post-processing operations like hashing and visualization. Experiments on the IJB-A dataset show that the proposed algorithm outperforms state of the art methods in verification and identification metrics, while requiring less training time.