MULTIMODAL APPROACH FOR FACE DETECTION AND VERIFICATION

Abstract

The availability of large annotated datasets and affordable computational power have led to impressive improvements in the performance of Convolutional Neural Networks (CNNs) on various face analysis tasks. In this paper, we describe a deep learning pipeline for unconstrained face identification and verification which achieves state-of-the-art performance on several benchmark datasets. We provide the design details of the various modules involved in automatic face recognition: face detection, landmark localization and alignment, and face identification/verification. We propose a novel face detector, Deep Pyramid Single Shot Face Detector (DPSSD), which is fast and detects faces with large scale variations (especially tiny faces). Additionally, we propose a new loss function, called the Crystal Loss, for the tasks of face verification and identification. Crystal Loss restricts the feature descriptors to lie on a hypersphere of a fixed radius, thus minimizing the angular distance between positive subject pairs and maximizing the angular distance between negative subject pairs. We provide evaluation results of the proposed face detector on challenging unconstrained face detection datasets. Then, we present experimental results for end-to-end face verification and identification on IARPA Janus Benchmarks A, B and C (IJB-A, IJB-B, IJB-C), and the Janus Challenge Set 5 (CS5).