tntrung / sdm_face_alignment Goto Github PK

View Code? Open in Web Editor NEW

99.0 11.0 71.0 430.04 MB

The Matlab implementation of Supervised Descent Method (SDM) for Face Alignment.

License: GNU General Public License v2.0

MATLAB 99.79% Mercury 0.21%

sdm_face_alignment's Introduction

Matlab Implementation of Supervised Descent Method

A simple Matlab implementation of Supervised Descent Method (SDM) for Face Alignment.

I provide both training and testing modules and one trained model of LFPW subset of 300-W dataset.

You can find the ogirinal paper of my implementation:

Xiong et F. De la Torre, Supervised Descent Method and its Applications to Face Alignment, CVPR 2013.

===========================================================================

Dependency:

Vlfeat library: http://www.vlfeat.org/
libLinear: www.csie.ntu.edu.tw/~cjlin/liblinear/

Datasets in use:

[300-W] http://ibug.doc.ic.ac.uk/resources/facial-point-annotations/

How to use:

Download 300-W data (i.e. LFPW) from above link and put into "./data" folder, then correct the dataset path to your dataset foler in setup.m

mkdir -p data

For example:

options.trainingImageDataPath = './data/lfpw/trainset/';

options.trainingTruthDataPath = './data/lfpw/trainset/';

options.testingImageDataPath = './data/lfpw/testset/';

options.testingTruthDataPath = './data/lfpw/testset/';
Download and install dependencies: libLinear, Vlfeat, mexopencv, put into "./lib" folder and compile if necessary. Make sure you already addpath(...) all folders in matlab. Check and correct the library path in setup.m.

mkdir -p lib

libLinear:
- Open Matlab
- Go to i.e. lib/liblinear-1.96/matlab/ in Matlab editor.
- Run make.m to comile *.mex files.
Vlfeat:
- cd lib/vlfeat/ && make
- cd ./toolbox in Matlab editor.
- Run vl_setup
- Compile mex Hog functions:
  
  cd misc mex -L../../bin/glnx86 -lvl -I../ -I../../ vl_hog.c
- Setup libvl.so path.
- Assume that your libvl.so located at: <vlfeat_folder>/bin/glnx86 Create soft link:
  
  ln -s <vlfeat_folder>/bin/glnx86/libvl.so /usr/local/libvl.so Check if the libvl.so is ready to use. ldd vl_hog.mexglx If libvl.so still not found. Add /usr/local/lib into /etc/ld.so.conf (sudo). sudo ldconfig ldconfig -p | grep libvl.so Check again: >> ldd vl_hog.mexglx
If you run first time. You should set these following parameters to learn shape and variation. For later time, reset to 0.

options.learningShape = 1; options.learningVariation = 1;
Do training:

run_training();
Do testing:

do_testing();

Note: in the program, we provide training models of LFPW (68 landmarks) in folder: "./model". The program does not optimize speed and memory during training, the memory problem may happens if you train on too much data.

sdm_face_alignment's People

Contributors

Stargazers

Watchers

sdm_face_alignment's Issues

Error

Hi, when I Run make.m to comile *.mex files, it give error:
Error: Could not find the compiler "cl" on the DOS path.
Use mex -setup to configure your environment properly.

E:\MATLAM\BIN\MEX.PL: Error: Unable to locate compiler.

Error: E:\matlam\toolbox\matlab\general\mex.m failed (line 222)
Unable to complete successfully.
=> Please check README for detailed instructions.

what's the problem?
thanks

How to test the code with Video files

I have trained the model and it is working correctly.

But, now I would like to detect face landmarks with videos. Is that possible?

Thank you.

do_learn_shape And do_learn_variation functions

Hi, tntrung. Thanks for your implementation.
Actually, I wonder why we need do_learn_shape and do_learn_variation functions. And after we get ShapeModel and DataVariation, what do we do with it?
Sorry for my silly question.

How to get the pts file and it includes 68 feature points.

Hi !
I'm sorry to bother you, but there is a problem with your code we have some questions confused me for a long time. I carefully read your code, I found your database has a PTS file, it contains 68 feature points. I've been thinking about when there is no the PTS file how can we run?
I do the experiment with their own data but i haven't a result, the reason is that I can't get a PTS file. I tried to manually tag within the coordinates of facial feature points on the PTS files, but the pictures too much, I a person cannot complete such a big task. How do you know the PTS file of each face image automatically get？

How to test images without ground-truth points?

In your codes, I saw in test phase, you used ground-truth points to generate initial points, but in practice, the test images have no ground-truth points, how do I test them?

Can you share your trained model?

Hi tntrung, I am runing do_testing, But some models are lost, so Can you share these models ti me?
Thank you!

calculate PTS File (.pts)

Hi,
I want to detect The 68 facial feature points for a new dataset for example feret. How can i calculate PTS File (.pts) for new images?
Thank you

mex vl_hog.c error

Hi, when I do step "mex -L../../bin/glnx86 -lvl -I../ -I../../ vl_hog.c", there occurs an error,
Error using mex
In file included from ../../vl/generic.h:18:0,
from ../mexutils.h:18,
from
/home/huneng/source/impSDM/lib/vlfeat/toolbox/misc/vl_hog.c:14:
../../vl/host.h:312:1: error:
expected identifier or ‘(’ before ‘/’ token
Is there any dependencies I haven't done? Thanks.

How to get the xx_sift.mexw32?

Hi！
I have a problem which is how to get the xx_sift.mexw32? My system is 32bits.

Training Error

Hi, thank you for your implementation, but I get error while training as bellow. I did every thing as written in readme.md file.

...
Stage: 1 - Image: 224
Stage: 1 - Image: 225
Subscripted assignment dimension mismatch.

Error in local_descriptors (line 55)
desc(ipts,:) = hog(im,xy(ipts,:),dsize);

Error in learn_single_regressor (line 160)
tmp = local_descriptors( cropIm_scale, ...

Error in run_training (line 42)
[R,new_init_shape,rms(icascade)] = learn_single_regressor( ...