In the OpenCV contrib modules we find 3 methods to work with face recognition:
recognizer = cv2.face.createEigenFaceRecognizer()
recognizer = cv2.face.createLBPHFaceRecognizer()
recognizer = cv2.face.createFisherFaceRecognizer()
but when I try to predict some face image, each of them returns a different confidence value.
nbr_predicted, conf = recognizer.predict( gray_image )
For example: for the same image, LBPH returns a confidence of 80 and the EigenFace returns a confidence of 1500.
I would like to know if each of them has some range in the confidence value?
For example, the LBPH method returns a confidence between 0 and 250, and the EigenFace returns a confidence between 0 and 3000.
I would like to know this so I can normalize the confidence values to compare the confidence results of each method.