The faces database is a set of 1024x768 pixels images which contain frontal faces in various sizes, locations, skin colors, races, etc. Each frame includes one image with no faces for comparison.
The database is free for academic purposes. Please acknowldege the use of this databse by citing this paper:
Cerf Moran, Harel Jonathan, Einhäuser Wolfgang, Koch Christof, "Predicting human gaze using low-level saliency combined with face detection", Advances in Neural Information Processing Systems (NIPS), 2007. PDF
For a list of the images used in each phase of the NIPS experiment, click here.
For more related papers see publications.
The database is broken into 3 categories:
1. Faces (229 images) - a set of images with faces. (jpg, tif)
2. Black and White (50 images) - a set of black and white images with faces. (jpg, tif)
3. Normalized exposures (24 images) - a set of images containing faces, with normalized exposures, for better comparison with saliency models that are based on luminosity and contrast measures. (jpg, tif)
Database Samples
|
| |
|
|