An Open Source Framework for Standardized Comparisons of Face Recognition Algorithms

Abstract

Abstract. In this paper we introduce the facereclib, the first software library that allows to compare a variety of face recognition algorithms on most of the known facial image databases and that permits rapid prototyping of novel ideas and testing of meta-parameters of face recognition algorithms. The facereclib is built on the open source signal processing and machine learning library Bob. It uses well-specified face recognition protocols to ensure that results are comparable and reproducible. We show that the face recognition algorithms implemented in Bob as well as third party face recognition libraries can be used to run face recognition experiments within the framework of the facereclib. As a proof of concept, we execute four different state-of-the-art face recognition algorithms: local Gabor binary pattern histogram sequences (LGBPHS), Gabor graph comparisons with a Gabor phase based similarity measure, inter-session variability modeling (ISV) of DCT block features, and the linear discriminant analysis on two different color channels (LDA-IR) on two different databases: The Good, The Bad, & The Ugly, and the BANCA database, in all cases using their fixed protocols. The results show that there is not one face recognition algorithm that outperforms all others, but rather that the results are strongly dependent on the employed database.

Protected Facial Biometric Templates Based on Local Gabor Patterns and Adaptive Bloom Filters

“…In a fair benchmark among four state-of-the-art algorithms for face recognition established in [21], using the same databases and protocols, LGBPHS achieved a top performance. Furthermore, LGBPHS requires no training.…”

“…More specifically, the face recognition algorithm considered is implemented in the Facereclib [21], a library comprising several face verification algorithms and database interfaces, implemented over the more general Bob platform.…”

Face Recognition in Challenging Environments: An Experimental and Reproducible Research Survey

“…There have been attempts to foment reproducibility of research results in the biometric community with the release of public software [20,23,31,32] and datasets [15,16,33,34]. Various biometric communities organize open challenges [35,36], for which web-based solutions for data access and result posting are particularly attractive [37].…”

“…2 Some algorithms are taken from the CSU Face Recognition Resources, 3 which provide the baseline algorithms for the Good, the Bad & the Ugly (GBU) face recognition challenge [21,22]. Finally, all experiments are executed using the FaceRecLib [23], 4 which offers an easy interface to run face recognition experiments either using already implemented face recognition algorithms, or rapidly prototyping novel ideas.…”

Biometric score calibration for forensic face recognition

“…The choice of these algorithms is based on their wide use and availability of standard implementations. Refer to [112] for details of the specific implementations of these algorithms. These stanadard implementations are available online here [113].…”