luckycallor/InsightFace-tensorflow

Question about: "face images should be well cropped here"

khanhnt opened this issue · 2 comments

Hi,

Thank you for your interesting work. I have tried to extract embedding with Pretrained Model, you mention that the face images should be well cropped here. Could you pls explain about it ? How well should I crop the faces ? Do I need to align the face ?

Thanks.

Have you solved this question?

Look at MTCNN or RetinaFace(deepinsight) face detections. They crop faces very tightly(Above eye brows to your chin, ears not included). Looking at Dlib you will see that after dlib face detection dlib face recognition loosens the bounding box of heads whereas this won't be required here.