zhzhanp / tcdcn-face-alignment Goto Github PK

View Code? Open in Web Editor NEW

229.0 13.0 89.0 2.4 MB

Matlab implementation of facial landmark detection by deep multi-task learning

MATLAB 100.00%

face-alignment matlab

tcdcn-face-alignment's Introduction

TCDCN face alignment tool

Matlab implementation of facial landmark detection by deep multi-task learning.

For a Win32 binary version (faster), please go to: http://mmlab.ie.cuhk.edu.hk/projects/TCDCN.html

Face alignment requires the bounding box from an external face detector. You may find a useful one from here: https://github.com/kpzhang93/MTCNN_face_detection_alignment.git

The provided tool is a re-implementation of the papers:

[1] Zhanpeng Zhang, Ping Luo, Chen Change Loy, Xiaoou Tang. Facial Landmark Detection by Deep Multi-task Learning, in Proceedings of European Conference on Computer Vision (ECCV), 2014

[2] Zhanpeng Zhang, Ping Luo, Chen Change Loy, Xiaoou Tang. Learning Deep Representation for Face Alignment
with Auxiliary Attributes. to appear in IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI).

tcdcn-face-alignment's People

Contributors

Stargazers

Watchers

Forkers

nanyomy akumar14 lijian8 gjtjx michaelxin cmxnono ltyscu geshiming lvchigo linzhineng caomw walkoncross zhiwenshao duhaoze11 ilovecv fanghaizhao ofirbb amos-zq milestonesvn paseam ohhhyeahhh tfy1028 zandaya aliscifp dan1900 j0k bygreencn zengjianping zymcool wshenx jwgu kissyzhou caozhengquan pinglmlcv bowrein dengcy028 dboyliao soledad89 geogreff jwmneu yiliangnie coocoky yang-fei jassonvia hsd315 boosting arasharchor elecun guoyijie tpys xpsunshine yangtairen francisyizhang zjucsxxd abc1225 nicoleqwerty tanghy2016 december-boy tim-lee-cn sybil12 zhangxujinsh ganwy2017 jacoblee121 afcarl chengyiwen sandwichfish facexteam luckynote fangjyshanghai sysuzyq balancewing artificially-ai lunar-r kailushan tianzhix xrosliang ypy1 qiy zspdczx thailand88 cike883 tmxk2018 tom1997 liaw05 liustenfan roomdestroyer danielsun-ni revhz

tcdcn-face-alignment's Issues

Where is the bounding box information of the MTFL dataset?

The annotations are only landmarks and attributes but no bounding box information.

Is there training code with instructions available?

Thank you for sharing your nice work!

I would like to use your implementation in my research but it is essential for my purposes to train the model myself. Is there training code with instructions available?

Thank you :)

would you please share the .m code?

Unable To run Demo

Hi I'm getting following error

Undefined function 'bbApply' for input arguments of type 'char'.

Error in
/TCDCN/preProcessing.p>preProcessing
(line 30)
TCDCN/main.p>main (line 7)

Error in demo (line 48)
main(listFileName,modelFile,outputFileName)

How to implement convolutioan with filter weights unshared?

In the TCDCN paper, the filter weights are not spatially shared. I want to implement the network in caffe, but caffe seems to not support the locally shared or unshared weights. Can you share your implement details of this part? And, which deep learning toolbox do you use?

Is Multi-Task Facial Landmark (MTFL) dataset included in CelebA?

I would like to know if MTFL dataset has any joint part with Celeb A

68 facial feature points

Hi,
I want to detect The 68 facial feature points for different datasets that for example have 1000 images. How can i estimate a facial bounding box because each image want a different bounding box?

what is the meaning of your preProcessing and postProcessing?

@zhzhanp
I load you model.mat and get the weights.
(1)first I resize the grayimage to 60x60;
(2)normalized to zero-mean and unit-variance:
[ grayimage-mean(grayimage)]/std(grayimage)
(3)using your weights to extract feature and get landmarks: vector(136);
(4) map the landmarks: landmarks*60
But I get the wrong results. Can you check the process is right??
Thanks!!

Facial landmarks on output image shift away

I have a problem with the output pic when I run visualize.m. It seems the detected landmarks shift away...
Does anyone have this problem as well?

Where is the dataset?

Hello,
Your website directed me here? Could you please show me the URL of the dataset?
Thank you very much in advance,

Trung