We implement several data augmentation methods, include traditional image augmentation methods, such like ShearXY，TranslateXY，Rotate，AutoContrast，Invert，Equalize，Solarize，Posterize，Contrast，Color，Brightness，Sharpness，Cutout, and Addmask，Addglass to generate face images wearing glasses and masks.
The glass style is randomly chosed from four templates.
The mask style is randomly chosed from five templates and the scale controls the cover magnitude.
landmark_AddHeadband2Face.py is the code for generating face images wearing fake headscarves, the input is the image list and corresponding 68 points landamark list.