pyhowfar

pyhowfar copied to clipboard

• Pytorch version of 'How far are we from solving the 2D & 3D Face Alignment problem? (and a dataset of 230,000 3D facial landmarks)'

  For official torch7 version please refer to [[https://github.com/1adrianb/face-alignment-training][face-alignment-training]]

  This is a reinplement of training code for 2D-FAN and 3D-FAN decribed in "How far" paper. Please visit [[https://www.adrianbulat.com][author's webpage]] or [[https://arxiv.org/abs/1703.07332][arxiv]] for technical details.

  Thanks for bearpaw's excellent work on human pose estimation [[https://github.com/bearpaw/pytorch-pose][pytorch-pose]] . And in this project, I reused a branch of helper function from pytorch-pose.

  Pretrained models are available soon.

** Requirments

• Install the latest [[http://pytorch.org][PyTorch]], version 0.2.1 is fully supported and there is no further test on older version.

*** Packages

- [[https://www.scipy.org/][scipy]]
- [[https://pytorch.org][torchvision]]
- [[https://pypi.python.org/pypi/progress][progress]](optional) for better visualization.

** Setup

1. Clone the github repository and install all the dependencies mentiones above.

#+BEGIN_SRC bash git clone https://github.com/hzh8311/pyhowfar cd pyhowfar #+END_SRC

2. Download the 300W-LP dataset from the authors webpage. In order to train on your own data the dataloader.lua file needs to be adapted.

3. Download the 300W-LP annotations converted to t7 format by paper author from [[https://www.adrianbulat.com/downloads/FaceAlignment/landmarks.zip][here]], extract it and move the landmarks folder to the root of the 300W-LP dataset.

** Usage

In order to run the demo please download the required models available bellow and the associated data.

#+BEGIN_SRC bash python main.py #+END_SRC

In order to see all the available options please run:

#+BEGIN_SRC bash python main.py --help #+END_SRC

** What's different?

• Pythoner friendly and there is no need for .t7 format annotations
• Add 300-W-LP test set for validation.
• Followed the excatly same training procedure described in the paper (except binary network part).
• Add model evaluation in terms of Mean error, [email protected]
• TODO: add evaluation on test sets (300W, 300VW, AFLW2000-3D etc.).

** Citation

#+BEGIN_SRC bibtex @inproceedings{bulat2017far, title={How far are we from solving the 2D & 3D Face Alignment problem? (and a dataset of 230,000 3D facial landmarks)}, author={Bulat, Adrian and Tzimiropoulos, Georgios}, booktitle={International Conference on Computer Vision}, year={2017} } #+END_SRC