Visual Geometry Group – University of Oxford

Overview

This page contains the download links for the source code for computing the VGG-Face CNN descriptor, described in [1].

The VGG-Face CNN descriptors are computed using our CNN implementation based on the VGG-Very-Deep-16 CNN architecture as described in [1] and are evaluated on
the Labeled Faces in the Wild [2] and the YouTube Faces [3] dataset.

Additionally the code also contains our fast implementation of the DPM Face detector of [3] using the cascade DPM code of [4].

Details of how to crop the face given a detection can be found in vgg_face_matconvnet package below in class faceCrop in +lib/+face_proc directory.

These models can be used for non-commercial research purposes under Creative Commons Attribution License.

Results

Downloads

• vgg_face_matconvnet.tar.gz: Face detection and VGG Face descriptor source code and models (MatConvNet)
• vgg_face_torch.tar.gz: VGG Face descriptor source code and models (Torch)
• vgg_face_caffe.tar.gz: VGG Face descriptor source code and models (Caffe)

Relevant Publications

[1]
O. M. Parkhi,
A. Vedaldi,
A. Zisserman

Deep Face Recognition

British Machine Vision Conference, 2015

[2]
G. B. Huang, M. Ramesh, T. Berg, E. Learned-Miller

Labeled faces in the wild: A database for studying face recognition in unconstrained environments.

Technical Report 07-49, University of Massachusetts, Amherst, 2007.

[3]
L. Wolf, T. Hassner, I. Maoz

Face Recognition in Unconstrained Videos with Matched Background Similarity.

Computer Vision and Pattern Recognition (CVPR), 2011.

[4]
M. Mathias, R. Benenson, M. Pedersoli, L. Van Gool

Face detection without bells and whistles.

European Conference on Computer Vision, 2014.

[5]
P. Felzenszwalb, R. Girshick, D. McAllester

Cascade Object Detection with Deformable Part Models.

Computer Vision and Pattern Recognition (CVPR), 2010.