MUCT Details
This page describes the camera setup and some other details of the
MUCT face database.
Camera views
Each subject was photographed with the five webcams shown below.
Camera a was directly in front of the subject's face, up to
variations in height of seated subjects.
No cameras were located to the left of the subject, since
those views can be approximated for model-building purposes by mirroring the images from the
cameras on the right.
Lighting sets
Ten different lighting setups were used, and each subject was
photographed (with all five webcams)
with two or three of these lighting sets.
Not every subject was shot with every lighting set,
to achieve diversity without too many images.
The following table gives details.
The full set of images for subject 000, for example, is (in low-resolution):
Manual landmarks
Each face is carefully manually landmarked with
76 landmarks
as shown on the right.
Filenames
The image filenames have the form i000qa-fn.jpg
where
- i prefixes all MUCT filenames
- 000 is the subject number
- q is the lighting set (q to z, see the above table)
- a is the camera view (a to e, see the above figure)
- f is f for female, m for male
- n is n for no glasses, g for glasses.
This naming scheme was chosen to make it easy to select subsets using file wildcards.
Contact
For more details, please feel free to contact milbo AT sonic PERIOD net.
Back to MUCT Homepage