Face key point prediction and normalization

1.EVPface normalization in

EVPThe key points of the predictions are in0～1之间

 所以,If you want to visualize,like zooming in

Just follow all the key pointscanves的width,和height进行等比例放大

将所有的landmarks都乘以width,

landmarks=np.reshape(lm,(-1,2))*256

然后可视化显示

 

 2.使用face_alignment框架

使用face_alignment框架,输入一张图片,What is returned is the point that is zoomed in according to the image