import torch
torch.backends.cudnn.benchmark = True
fntr.eval_epoch()
if __name__ == '__main__':
# Disable traceback on Ctrl+c
import sys
import signal
signal.signal(signal.SIGINT, lambda x, y: sys.exit(0))
import configargparse
import numpy as np
np.set_printoptions(linewidth=np.inf)
parser = configargparse.ArgParser()
aae_training.add_arguments(parser)
# Dataset
parser.add_argument('--dataset', default=['w300'], type=str, choices=cfg.get_registered_dataset_names(),
nargs='+', help='dataset for training and testing')
parser.add_argument('--test-split', default='full', type=str, help='test set split for 300W/AFLW/WFLW',
choices=['challenging', 'common', '300w', 'full', 'frontal']+wflw.SUBSETS)
parser.add_argument('--benchmark', default=False, action='store_true', help='evaluate performance on testset')
# Landmarks
parser.add_argument('--sigma', default=7, type=float, help='size of landmarks in heatmap')
parser.add_argument('--ocular-norm', default=lmconfig.LANDMARK_OCULAR_NORM, type=str,
help='how to normalize landmark errors', choices=['pupil', 'outer', 'none'])
args = parser.parse_args()
import configargparse
np.set_printoptions(linewidth=np.inf)
# Disable traceback on Ctrl+c
import signal
signal.signal(signal.SIGINT, lambda x, y: sys.exit(0))
parser = configargparse.ArgParser()
defaults = {
"batchsize": 50,
"train_encoder": False,
"train_decoder": False
}
aae_training.add_arguments(parser, defaults)
# Dataset
parser.add_argument(
"--dataset",
default=["w300"],
type=str,
help="dataset for training and testing",
choices=["rhpe", "rsna"],
nargs="+",
)
# Landmarks
parser.add_argument(
"--lr-heatmaps",
default=0.001,