资源算法Intensity-Depth-Face-Alignment

Intensity-Depth-Face-Alignment

2020-03-25 | |  35 |   0 |   0

Intensity-Depth Face Alignment

This is the source code (with dataset) for my ICONIP2015 paper "Intensity-Depth Face Alignment Using Cascade Shape Regression".

This code is largely based on my FaceX project. Therefore, if you have any problem related to the code itself (such as compiling and label format), please refer to https://github.com/delphifirst/FaceX/blob/master/README.md first.

How To Repeat My Result

First, you can compile the testing code (this part of code is in FaceX directory), and use the pre-trained models to repeat my result. Make sure your current directory is Intensity-Depth-Face-AlignmentFaceX, and run the command like this:

..x64ReleaseFaceX.exe pre-trained_modelsmodel_0.8.xml.gz

The program will open this model (the number "0.8" in the model name means that feature ratio is 80% intensity and 20% depth), and use the testing data in Intensity-Depth-Face-AlignmentFaceXtest to give you a result.

Then, if you want, you can train your own model. Compile the training code (this part of code is in FaceX-Train directory), make sure your current directory is Intensity-Depth-Face-AlignmentFaceX-Train, and then run the command like this:

..x64ReleaseFaceX-Train.exe config.txt

The program will open config.txt, use the settings in this file, and use the training data in Intensity-Depth-Face-AlignmentFaceX-Traintrain to train a model for you. You can tune parameters in config.txt. For my paper, the most important parameters are Rho and Alpha.


上一篇:FaceAlignmentJNI

下一篇:face_alignment_cvpr_2012

用户评价
全部评价

热门资源

  • seetafaceJNI

    项目介绍 基于中科院seetaface2进行封装的JAVA...

  • spark-corenlp

    This package wraps Stanford CoreNLP annotators ...

  • Keras-ResNeXt

    Keras ResNeXt Implementation of ResNeXt models...

  • capsnet-with-caps...

    CapsNet with capsule-wise convolution Project ...

  • shih-styletransfer

    shih-styletransfer Code from Style Transfer ...