CMU Pose, Illumination, and Expression (PIE) Database - Terence Sim, Simon Baker, and Maan Bsat

Aug 27, 2026

Upload by Levi

A database of over 40,000 facial images was collected between October 2000 and December 2000 from 68 people to support face analysis under varying conditions. Each subject was imaged in 13 poses, under 43 illumination conditions, and with 4 expressions (neutral, smile, blink, and talk). The paper explains the imaging hardware in the CMU 3D Room, the flash system used for illumination augmentation, how the data are organized, potential uses, and how to obtain the database.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages6

Reading Time15 min

AccessFree to download

Abstract

Introduction

Capture Apparatus and Procedure

Setup of the Cameras: Pose

Appeared in the 2002 International Conference on Automatic Face and Gesture Recognition The CMU Pose, Illumination, and Expression (PIE) Database Terence Sim, Simon Baker, and Maan Bsat The Robotics Institute, Carnegie Mellon University 5000 Forbes Avenue, Pittsburgh, PA 15213 Abstract Between October 2000 and December 2000 we collected a database of over 40,000 facial images of 68 people. Using the CMU 3D Room we imaged each person across 13 different poses, under 43 different illumination conditions, and with 4 different expressions. We call this database the CMU Pose, Illumination, and Expression (PIE) database. In this paper we describe the imaging hardware, the collection procedure, the organization of the database, several potential uses of the database, and how to obtain the database. 1 Introduction People look very different depending on a number of factors. Perhaps the three most signiﬁcant factors are: (1) the pose; i.e. the angle at which you look at them,(2) the illumination conditions at the time, and (3) their facial expression; i.e. whether or not they are smiling, etc. Although several other face databases exist with a large number of subjects [Philips et al., 1997], and with signiﬁcant pose and illumination variation [Georghiades et al., 2000], we felt that there was still a need for a database consisting of a fairly large number of subjects, each imaged a large number of times, from several different poses, under signiﬁcant illumination variation, and with a variety of facial expressions. Between October 2000 and December 2000 we collected such a database consisting of over 40,000 images of 68 subjects. (The total size of the database is about 40GB.) We call this database the CMU Pose, Illumination, and Expression (PIE) database. To obtain a wide variation across pose, we used 13 cameras in the CMU 3D Room [Kanade et al., 1998]. To obtain signiﬁcant illumination variation we augmented the 3D Room with a “ﬂash system” similar to the one constructed by Athinodoros Georghiades, Peter Belhumeur, and David Kriegman at Yale University [Georghiades et al., 2000]. We built a similar system with 21 ﬂashes. Since we captured images with, and without, background lighting, we obtained 21 􀀂 2 + 1 = 43 different illumination conditions. Finally, we asked the subjects to pose with several different “expressions.” In particular, we asked them to give a neutral expression, to smile, to blink (i.e. shut their eyes), and to talk. These are probably the four most frequently occurring “expressions” in everyday life. Figure 1: The setup in the CMU 3D Room [ Kanade et al., 1998] . The subject sits in a chair with his head in a ﬁxed position. We used 13 Sony DXC 9000 (3 CCD, progressive scan) cameras with all gain and gamma correction turned off. We augmented the 3D Room with 21 Minolta 220X ﬂashes controlled by an Advantech PCL-734 digital output board, duplicating the Yale “ﬂash dome”used to capture the database in [Georghiades et al., 2000] . Capturing images of every person under every possible combination of pose, illumination, and expression was not practical because of the huge amount of storage space required. The PIE database therefore consists of two major partitions, the ﬁrst with pose and illumination variation, the second with pose and expression variation. There is no simultaneous variation in illumination and expression because it is more difﬁcult to systematically vary the illumination while a person is exhibiting a dynamic expression. In the remainder of this paper we describe the capture hardware in the CMU 3D Room, the capture procedure, the organization of the database, several possible uses of the database, and how to obtain a copy of it. 2 Capture Apparatus and Procedure 2.1 Setup of the Cameras: Pose Obtaining images of a person from multiple poses requires either multiple cameras capturing images simultaneously, or multiple “shots” taken consecutively (or a combination of the two.) T

FAQs

What data does the CMU PIE database contain?

It contains over 40,000 facial images from 68 people captured across 13 poses, 43 illumination conditions, and 4 expressions.

How was illumination variation achieved?

Illumination was augmented in the CMU 3D Room using a flash system with 21 flashes, producing 43 illumination conditions including captures with and without background lighting.

Why does the database avoid simultaneous variation of illumination and expression?

Because systematically varying illumination while subjects exhibit dynamic expressions is more difficult, so the dataset uses two main partitions instead.

Aug 27, 2026

Upload by Levi