type=str,
help='train/train_dev/test/train_k_fold/test_k_fold')
parser.add_argument('--cuda', default='3', help='CUDA device ID', type=str)
parser.add_argument(
'--date',
default=None,
type=str,
help='please specify the running date when test one model')
parser.add_argument('--nocsv',
action='store_true',
help="don't store a csv file to the desktop")
args = parser.parse_args()
os.environ["CUDA_VISIBLE_DEVICES"] = args.cuda
csv_date = time.strftime("%F", time.localtime())
csv_helper = CSVHelper(save_dir='~/Desktop/ZL-baseline/',
csv_date=csv_date,
train_mode=args.mode,
append_mode=True)
cfg = Configurer(args.config_path, verbose=True)
train_metrics_list = []
dev_metrics_list = []
test_metrics_list = []
assert cfg.DATASET.fabrics == 'groups'
assert args.mode in ['train', 'train_k_fold']
cfg.DATASET.fabrics = range(1, 6)
for g_id in cfg.DATASET.fabrics:
cfg.DATASET.fabric = 'group{0}'.format(g_id)
print('{0:%^64}'.format(cfg.DATASET.fabric))
if g_id == 5: cfg.DATASET.color_mode = 'gray'
model = UNet(cfg=cfg)
trainer = TorchTrainer(cfg=cfg, core_model=model)
ds_builder = MaskDSB(cfg=cfg)
type=str,
help='train/train_dev/test/train_k_fold/test_k_fold')
parser.add_argument('--cuda', default='3', help='CUDA device ID', type=str)
parser.add_argument(
'--date',
default=None,
type=str,
help='please specify the running date when test one model')
parser.add_argument('--nocsv',
action='store_true',
help="don't store a csv file to the desktop")
args = parser.parse_args()
os.environ["CUDA_VISIBLE_DEVICES"] = args.cuda
csv_date = time.strftime("%F", time.localtime())
csv_helper = CSVHelper(save_dir='~/Desktop/ZL-baseline/',
csv_date=csv_date,
train_mode=args.mode,
append_mode=True)
cfg = Configurer(args.config_path, verbose=True)
train_metrics_list = []
dev_metrics_list = []
test_metrics_list = []
fabrics = cfg.DATASET.fabrics
if fabrics in ['all', 'groups', 'total']:
cfg.DATASET.fabrics = range(1, 20)
else:
raise NotImplementedError
for p_id in cfg.DATASET.fabrics:
cfg.DATASET.fabric = 'pattern{0}'.format(p_id)
print('{0:%^64}'.format(cfg.DATASET.fabric))
model = UNet(cfg=cfg)
trainer = TorchTrainer(cfg=cfg, core_model=model)