The CMU Pose, Illumination, and Expression (PIE) database was collected in the fall of 2000, containing over 40,000 facial images of 68 subjects. The database was captured using the CMU 3D Room with 13 cameras, 21 flashes, and various illumination conditions. Each subject was imaged in 13 different poses, under 43 different illumination conditions, and with 4 different expressions. The database includes color images of size 640×486, with 600 images per subject, totaling around 40GB. The database was created to provide a large number of images with significant variation in pose, illumination, and expression, which can be used for evaluating the robustness of face recognition algorithms. The 13 cameras were positioned in an arc around the subject, capturing images from various angles. The 21 flashes provided different illumination conditions, with images captured both with and without background lighting. The database also includes metadata such as head and camera locations, background images, color calibration images, and personal attributes of the subjects. The PIE database has a wide range of potential applications, including evaluating pose-invariant face detectors, head pose estimation algorithms, and the robustness of face recognition algorithms to illumination, expression, and pose. It can also be used for 3D face model building using stereo or photometric stereo techniques. The database is available for distribution to researchers worldwide, and can be obtained by contacting the second author or visiting the PIE database website. The database was funded by the U.S. Office of Naval Research.The CMU Pose, Illumination, and Expression (PIE) database was collected in the fall of 2000, containing over 40,000 facial images of 68 subjects. The database was captured using the CMU 3D Room with 13 cameras, 21 flashes, and various illumination conditions. Each subject was imaged in 13 different poses, under 43 different illumination conditions, and with 4 different expressions. The database includes color images of size 640×486, with 600 images per subject, totaling around 40GB. The database was created to provide a large number of images with significant variation in pose, illumination, and expression, which can be used for evaluating the robustness of face recognition algorithms. The 13 cameras were positioned in an arc around the subject, capturing images from various angles. The 21 flashes provided different illumination conditions, with images captured both with and without background lighting. The database also includes metadata such as head and camera locations, background images, color calibration images, and personal attributes of the subjects. The PIE database has a wide range of potential applications, including evaluating pose-invariant face detectors, head pose estimation algorithms, and the robustness of face recognition algorithms to illumination, expression, and pose. It can also be used for 3D face model building using stereo or photometric stereo techniques. The database is available for distribution to researchers worldwide, and can be obtained by contacting the second author or visiting the PIE database website. The database was funded by the U.S. Office of Naval Research.