how much size of images face detected & what is minimum and maximum size of images supported ( like KB, MB ) and also supported for which resolution area ( Width x Hight)
all models take image of any size and convert it internally to a square image with per-model fixed size
(if input image is not square, it will be extended with borders to achieve perfect square)
face detection: 512px
then crop each detected face and resize it to 112px to run landmarks, expressions and descriptors
age and gender model takes same cropped face, but resizes to 150px
how much size of images face detected & what is minimum and maximum size of images supported ( like KB, MB ) and also supported for which resolution area ( Width x Hight)