python - Dlib - fixed size of ROI in face detect

Question

I have a problem with face detection with the dlib library. I use the landmask coordinates to identify the forehead area. The problem is that I need that the size of the ROI is always the same, and instead unfortunately frame by frame the size of the ROI changes. Is there a way I can take the roi in the first frame and use that size for all the other frames?

self.faces = self.detector(self.frame_in, 0)

        if len(self.faces) == 1:
            self.face_points = self.predictor(self.frame_in, self.faces[0])

            if self.Start:

                roi,self.fh_left, self.fh_right, self.fh_top, self.fh_bottom = self.get_roi(self.frame_in,self.face_points)  

in the last instruction get the extracted roi and also the coordinates. I would like the roi size to remain unchanged while parsing all other frames.

question from:https://stackoverflow.com/questions/66055063/dlib-fixed-size-of-roi-in-face-detect

Answer 1

Waitting for answers