Face Detect and Removal in Preprossers.

[elpollodiablo1]

Don't know if this is in the right spot....... while using lineart_realistic or other highly detailed preprocessors, the face has good details, but this impacts the created image greatly. Which is good if you want it to have the same facial features as the original image.

I did some testing and found out if you remove the face from lineart preprocessed images it allows for more freedom for facial featrues that are in model you are using. I wrote up a little script useing standard 4 point face detection using face_recognition python library and just deleted out the rectangle/square face detection region. That probably isnt the best implementation of face detect.

It looks like face detect is already built into a few preprocessors(openpose face, midiapipe_face) but I don't see the capability in lineart_realistic. What would be the best way to create a new preprocessor for lineart(possible HED, pidinet, canny) that would detect faces and remove them?