tiqq111 / 2d-face-landmark-localization

2D face landmark localization with PyTorch

2D-face-landmark-localization

This repository contains the code of 2D face landmark localization with PyTorch

To run this code, you need to have following libraries:

• OpenCV - We tested our code with version > 3.0;
• PyTorch - We tested with v0.4.0;
• Numpy- We tested with 1.10.0;
• Python - We tested our code with Pythonv3. If you are using Python v2, please feel free to make necessary changes to the code;

Training, evaluate and export.

Before training started, make sure the following requirements.

• Training dataset and Test dataset must should be crop the face with some face detection algorithm such as dlib and so on.
• We should have a file list which contains the image name and landmark's coordinates such as: face_01.jpg, 1,2,3,4,5,6 ...
• You should argument you dataset with scael, mirror, affine, and I have put those code in the file of data_prepare.

The following example shows how to train the code:

bash
# From the repo's root directory
# Training the code
sh run.sh

bash
# From the repo's root directory
# test the model
sh eval.sh

The result

License

This project is licensed under the MIT License - see the LICENSE.md file for details