1:需要人臉的均值圖片
2:需要人臉的mean_shape
3: 需要人臉平均的128*128的均值和方差;之後採用減去均值除以方差;
4:兩階段的需要旋轉和對齊;
5:添加預訓練模型
論文:RetinaFace: Single-stage Dense Face Localisation in the Wild 代碼:https://github.com/deepinsight/insightface/tree/ma
論文:CenterFace: Joint Face Detection and Alignment Using Face as Point Github:https://github.com/Star-Clouds/CenterFace