def recognize_face(recognizer, image, face_cascade, eye_cascade, face_size, threshold):
found_faces = []
gray, faces = detect_faces(image, face_cascade, eye_cascade, return_gray=1)
# If faces are found, try to recognize them
for ((x, y, w, h), eyedim) in faces:
label, confidence = recognizer.predict(cv2.resize(level_face(gray, ((x, y, w, h), eyedim)), face_size))
if confidence < threshold:
found_faces.append((label, confidence, (x, y, w, h)))
return found_faces
def _extract_faces(a_dir, folder, do_level_face=False):
faceCascade = cv2.CascadeClassifier(config.FACE_CASCADE_FILE)
eyeCascade = cv2.CascadeClassifier(config.EYE_CASCADE_FILE)
the_path = join(a_dir, folder)
result = []
for img in [f for f in os.listdir(the_path) if _supported_img(f)]:
img_path = join(the_path, img)
image, faces = detect_faces(cv2.imread(img_path), faceCascade, eyeCascade, True)
if len(faces) == 0:
print("No face found in " + img_path)
for ((x, y, w, h), eyedim) in faces:
if not do_level_face:
result.append(image[y:y+h, x:x+w])
else:
result.append(level_face(image, ((x, y, w, h), eyedim)))
#result.append(image[y:y+h, x:x+w])
return result