face-alignment-with-68-point

Requirement

caffe & python2

1.Download the 300W dataset and run create_raw_300W_equal_bbox_train.py to obtain the annotation.

2.Prepare the dataset using generate_h5_300w_scale_V4.py ; This performs data augmentation and generates the input for the Caffe.

3.cd train_300w_68_vanilla and Run train.sh

you can download the 300W from here： https://github.com/FaceGg/DataSets

Name		Name	Last commit message	Last commit date
Latest commit History 20 Commits
Model_68Point		Model_68Point
common		common
dataset/300W		dataset/300W
train_300w_68_vanilla		train_300w_68_vanilla
.gitattributes		.gitattributes
.gitignore		.gitignore
README.md		README.md
config_300w.txt		config_300w.txt
generate_h5_300w_scale_V4.py		generate_h5_300w_scale_V4.py
predict_vanilla_fd_one.py		predict_vanilla_fd_one.py