The FacePix project
Center for Cognitive Ubiquitous Computing (CUbiC)
Arizona State University

Contact: <john.black@asu.edu>

Our motivation for developing FacePix

In the past, face databases such as FERET, XM2VTS, the CMU PIE Database, the Yale Face Database, the Yale B Database, and the MIT Database have been popular for evaluating the performance of face recognition algorithms. These databases provide face images captured with a variety of pose angles and illumination angles. However, they do not include sets of face images taken across a finely incremented spectrum of precisely calibrated pose angles, or illumination angles. Thus, researchers have not been able to (1) precisely measure the robustness of their face detection or recognition algorithms with respect to incremental variations in pose angle or illumination angle, or (2) compare the performance of their face detection or recognition algorithms to the algorithms proposed by other researchers.

Our motivations for developing the FacePix database are (1) to evaluate our own research aimed at developing robust face detection and recognition, and (2) to provide a face database that allows researchers to perform precise measurements of robustness and relative performance with respect to variation in pose angle and illumination. The development of the FacePix(30) database is the first step in an ongoing effort to provide comprehensive, well calibrated face image datasets with a wide spectrum of pose angles and illumination angles. The ultimate aim of this project is to support and motivate the development of face detection, and face recognition algorithms that are more robust with respect to pose angle and illumination angle.

What does this document contain?

This document contains:
  • A description of the content of the FacePix(30) database
  • The naming conventions of the files within the FacePix(30) package
  • The organization of the files in the downloaded FacePix(30) package
  • An overview of how the FacePix(30) database was created
  • The FacePix(30) license agreement