We provide the Pose-Aware CNN Models from the paper I. Masi, S. Rawls, G. Medioni, P. Natarajan "Pose-Aware Face Recognition in the Wild", in Proc. of CVPR 2016 [1] along with some more data.
In particular, we provide the following:
- PAMs CNN Models for AlexNet and VGGNet.
- Yaw estimates from our 3D pose estimation module on IJB-A [3].
You can try to generare your own rendered images using the code provided by [2].
Each CNN model is represented with the average image computed in the training set. For out-of-plane rotation, a 3D generic model is displayed to clarify that the alignment is 3D-aided but the CNN is trained only on RGB data.
Each model contains Caffe CNN parameters, a deploy file and the average face computed over the training set. Download the PAMs from the links below.
We provide our own pose estimation in IJB-A [3] dataset in terms of yaw angles. The tar can be downloaded here:
Download the yaw estimation in IJB-A (MATLAB)
The tar contains the yaw angles and a main MATLAB file to see how to use the data. Each image maps to a yaw angle using the filename and subject id as a key to index the image.
The key is built as <subjectID>_<filename> removing the extension from the filename and replacing '/' with '-'.
E.g. 1673,img/8580.jpg becomes 1673_img-8580
Please cite our paper with the following bibtex if you use our PAMs or our dataset:
@INPROCEEDINGS{masi2016cvpr,
author={Iacopo Masi and Stephen Rawls and G{\'e}rard Medioni and Prem Natarajan},
booktitle={CVPR},
title={Pose-{A}ware {F}ace {R}ecognition in the {W}ild},
year={2016}
}
[1] I. Masi, S. Rawls, G. Medioni, P. Natarajan "Pose-Aware Face Recognition in the Wild", CVPR 2016
[2] T. Hassner, S. Harel, E. Paz and R. Enbar "Effective Face Frontalization in Unconstrained Images", CVPR 2015
[3] Brendan F. Klare, Ben Klein, Emma Taborsky, Austin Blanton, Jordan Cheney, Kristen Allen, Patrick Grother, Alan Mah, Anil K. Jain, "Pushing the Frontiers of Unconstrained Face Detection and Recognition: IARPA Janus Benchmark A", CVPR 2015
- July 2016, First Release
The SOFTWARE PACKAGE provided in this page is provided "as is", without any guarantee made as to its suitability or fitness for any particular use. It may contain bugs, so use of this tool is at your own risk. We take no responsibility for any damage of any sort that may unintentionally be caused through its use.
If you have any questions, drop an email to iacopo.masi@usc.edu and srawls@isi.edu or leave a message below with GitHub (log-in is needed).
Thanks for sharing the models. Are these models eligible for commercial use?