On Face Detection and Performance Evaluation for Biometrics-based Face Systems

Mahmoud Hassaballah, Kenji Murakami, and Shun Ido

Keywords

Biometrics, Face detection, Performance Evaluation Measures, Golden Ratio

Abstract

Biometrics-based personal identification is considered as an effective method for automatically recognizing a persons identity. Face detection is a fundamental step in the biometrics-based face systems and many other applications from computer vision field. Most of the face-related applications such as face recognition and face tracking assume that the face region has been perfectly detected. To adapt a certain algorithm in these applications, evaluation of its performance is needed. Unfortunately, it is difficult to evaluate the performance of a face detection algorithm due to the lack of universal criteria in the literature. In this paper we introduce a new evaluation criterion for face detection algorithms. This criterion is based on the golden ratio definition of the perfect human face. The new evaluation criterion is simple and more realistic compared to the existing one. Using the proposed criterion five Harr-cascade classifiers provided by Intel's OpenCV have been quantitatively evaluated on three common databases and a new challenge dataset where the people wear head scarf to show their robustness and weakness as these classifiers have been never evaluated or compared before. After that a comparison between the best Harr-classifier and other face detection algorithms has been presented.

Important Links:

Go Back