This project is built by reproducing (at least partially) the face alignment algorithm in the CVPR 2014 paper:
Face Alignment at 3000 FPS via Regressing Local Binary Features. Shaoqing Ren, Xudong Cao, Yichen Wei, Jian Sun; The IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2014, pp. 1685-1692
How to run the codes?
(1) First of all, we need prepare datasets, such as afw, lfpw, helen, ibug, etc. All these can be downloaded freely from http://ibug.doc.ic.ac.uk/resources/facial-point-annotations.
(2) For training, run train_model.m with appropriate dataset name. Before that, define dbnames as {'Dataset_a', 'Dataset_b', ..., } for input.
(3) For testing, run test_model.m with dataset name dbnames and pre-trained model as input.
Dependencies
(1) liblinear: http://www.csie.ntu.edu.tw/~cjlin/liblinear/.
Learned Model
Off-the-shelf model can be downloaded here: http://pan.baidu.com/s/1i325Rbn, whose configure file can be found in folder "models". Its performance is analogy to the lbf_fast model evaluated in the original paper.
Q&A
Q_1: How to get the file Path_Images.txt?
A_1: It can be obtained by run bat file in the root folder of a dataset, the code is simply "dir /b/s/p/w *.jpg>Path_Images.txt".
Q_2: What is Ts_bbox.mat?
A_2: This problem is solved in recent version. Ts_bbox is a transformation matrix to adapt bounding boxes obtained from face detector to the boxes suitable for the face alignment algorithm.
Q_3: How to define the input variable dbnames in train_model and test_model functions?
A_3: It is formed as a cell array {'dbname_1' 'dbname_2' ... 'dbname_N'}. For example, if we use the images in afw for trainig, we then define it as {'afw'}.
Q_4: Why does an error occur when initializing parallel computing?
A_4: It may be caused by Matlab version. For Matlab 2014, it will be okay. For earlier version, please use the following commands:
if params.isparallel
if matlabpool('size') <= 0
matlabpool('open','local',4);
else
disp('Already initialized');
end
end
Q_5: Some function correspondences from Matlab 2014 to older version
A_5: fitgeotrans -> cp2tform, transformPointsForward -> tformfwd
For those Tecent QQ users, we can discuss more in the group: 180634020.