required=False,
help='1 to show region_of_interest otherwise 0, default is 0')
args = vars(ap.parse_args())
DEFAULT_PARAMS = {
'bounding_box': (True, bool),
'landmarks': (False, bool),
'delaunay_triangulation': (False, bool),
'region_of_interest': (False, bool),
}
for k in args:
if not k in ["input", "skip_rate"]:
args[k] = (DEFAULT_PARAMS[k][0]
if args[k] is None else utils.arg2bool(args[k]))
if k == "skip_rate":
args["skip_rate"] = (1 if args["skip_rate"] is None else int(
args["skip_rate"]))
# Opencv Haar
haar_detector = cv2.CascadeClassifier(
"face_detectors/haarcascade_frontalface.xml")
# Opencv LBP
lbp_detector = cv2.CascadeClassifier(
"face_detectors/lbpcascade_frontalface.xml")
# DLIB HoG
hog_detector = dlib.get_frontal_face_detector()
ap.add_argument(
'-he',
'--histogram_equalization',
required=False,
help='1 to apply histogram_equalization otherwise 0, default 0')
ap.add_argument('-ws',
'--window_size',
required=False,
help='Window size, comma separated values')
args = vars(ap.parse_args())
if args["histogram_equalization"] is None:
hist_eq = False
else:
hist_eq = utils.arg2bool(args["histogram_equalization"])
# DLIB HoG
hog_detector = dlib.get_frontal_face_detector()
# Opencv DNN
modelFile = "face_detectors/dnn_tf.pb"
configFile = "face_detectors/dnn_tf.pbtxt"
net = cv2.dnn.readNetFromTensorflow(modelFile, configFile)
conf_threshold = 0.7
shape_predictor = dlib.shape_predictor(
"face_detectors/shape_predictor_68_face_landmarks.dat")
def get_model_compatible_input(gray_frame, face):
args = vars(ap.parse_args())
DEFAULT_BOOLEAN_PARAMS = {
'shuffle': True,
'save_model': False,
'save_architecture': False,
'save_confusion_matrix': False,
'save_training_history': False,
}
for k in args:
if k in DEFAULT_BOOLEAN_PARAMS:
args[k] = (
DEFAULT_BOOLEAN_PARAMS[k]
if args[k] is None else
utils.arg2bool(args[k])
)
DEFAULT_NONBOOLEAN_PARAMS = {
'random_state': (42, int),
'train_ratio': (0.85, float),
'lr_scheduler': (None, str),
'early_stopping': (None, str),
'train_generator': (None, str),
'batch_size': (24, int),
'epochs': (50, int),
'optim': ("adam", str),
'learning_rate': (0.01, float),
}
for k in args: