QMUL Multiview Face Dataset
Queen Mary University
of London Multiview Face Dataset consists of automatically aligned, cropped and
normalised face images of 48 people, of which 37 people are in greyscale of 100x100
image size and 11 people in colour of 56x56 image size. Each person has 133
facial images covering a viewsphere of +/-90 degrees in yaw and +/-30 degrees
in tilt at 10 degrees increment. An example is shown below.
The dataset can be downloaded
from here
(40MB).
An example of tracking
face continously using a model built from this dataset is shown in this
video (2.3MB). More details are in J.
Sherrah and S. Gong. "Fusion of
Perceptual Cues for Robust Tracking of Head Pose and Position". Pattern
Recognition, Vol. 34, No. 8, pp. 1565-1572, 2001.
References:
1. S.
Gong, S. McKenna and A. Psarrou. Dynamic
Vision: From Images to Face Recognition, 364 pages,
2.
S. Gong, S.
McKenna, and J.J. Collins. "An Investigation
into Face Pose Distributions". In Proc. IEEE
International Conference on Automatic Face and Gesture Recognition, pp. 265-270,
3. S.
Gong, Eng-Jon Ong and S. McKenna. "Learning
to Associate Faces across Views in Vector Space of Similarities to Prototypes".
In Proc. British Machine
Vision Conference, pp. 54-63,