Skip to content

Implementation of PFLD A Practical Facial Landmark Detector , reference to https://arxiv.org/pdf/1902.10859.pdf

Notifications You must be signed in to change notification settings

myyang19770915/PFLD

 
 

Repository files navigation

PFLD

Table 1:
The code for cauculating euler angles prediction loss have been released.

Table 2:
I`ve done the job and fixed the memory leak bug: The code has a flaw that i calculate euler angles ground-truth while training process,so the training speed have slowed down because some work have to be finished on the cpu ,you should calculate the euler angles in the preprocess code

Table 3:It is an open surce program reference to https://arxiv.org/pdf/1902.10859.pdf , if you find any bugs or anything incorrect,you can notice it in the issues and pull request,be glad to receive you advices.
And thanks @lucknote for helping fixing existing bugs.

EASY TO TRAIN:

STEP1:data/SetPreparation.py
STEP2:train.sh

SAMPLE IMGS:

Image text Image text Image text Image text Image text Image text Image text

SAMPLE VIDEO:

Image text

CONTACT US:

If you have any questiones ,please contact us! Also join our QQ group(945933636) for more information.

About

Implementation of PFLD A Practical Facial Landmark Detector , reference to https://arxiv.org/pdf/1902.10859.pdf

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 99.3%
  • Shell 0.7%