Template-based Multi-Domain Face Recognition

TL;DR

A template generation algorithm called Norm Pooling is introduced and it is shown that it outperforms average pooling across different domains and networks, on the IARPA JANUS Benchmark Multi-domain Face (IJB-MDF) dataset.

Abstract

Despite the remarkable performance of deep neural networks for face detection and recognition tasks in the visible spectrum, their performance on more challenging non-visible domains is comparatively still lacking. While significant research has been done in the fields of domain adaptation and domain generalization, in this paper we tackle scenarios in which these methods have limited applicability owing to the lack of training data from target domains. We focus on the problem of single-source (visible) and multi-target (SWIR, long-range/remote, surveillance, and body-worn) face recognition task. We show through experiments that a good template generation algorithm becomes crucial as the complexity of the target domain increases. In this context, we introduce a template generation algorithm called Norm Pooling (and a variant known as Sparse Pooling) and show that it outperforms average pooling across different domains and networks, on the IARPA JANUS Benchmark Multi-domain Face (IJB-MDF) dataset.

Figures (3)

• Figure 1: A typical pipeline for generating a face template from a set of media (images/frames) corresponding to a subject. The template generation algorithm determines the coefficient $c_i$ for each media feature $\textbf{f}_i$. The resulting probe template corresponding to a subject is matched against the gallery templates to determine the identity of the subject. We generate one template per subject per domain.
• Figure 2: Face detection results on visible images captured at 300, 400, and 500m in the IJB-MDF dataset. Left: Original images. Right: Face detection results.
• Figure 3: Histograms of face detection probability scores in a template