def __init__(self, args, train=True):
super().__init__()
self.args = args
self.train = train
self.dir_hr = join(args.dir_datasets + '/DIV2K/HR')
self.dir_lr = [
join(args.dir_datasets + '/DIV2K/LR/X' + str(scale))
for scale in args.upscale
]
self.n_train = args.n_train
self.n_test = 20
if train:
self.images_hr = [
entry.path for entry in scandir(self.dir_hr)
if is_image_file(entry.name)
][:self.n_train]
else:
self.images_hr = [
entry.path for entry in scandir(self.dir_hr)
if is_image_file(entry.name)
][self.n_train:self.n_train + self.n_test]
self.images_lr = self._get_lr()
def __init__(self, args):
super().__init__()
self.args = args
self.dir_hr = join(args.dir_datasets, args.data_train + '/HR')
self.images_hr = [
entry.path for entry in scandir(self.dir_hr)
if is_image_file(entry.name)
]
def __init__(self, args):
super().__init__()
self.upscale = args.upscale
self.dir_hr = join(args.dir_datasets, args.data_test + '/HR')
self.dir_lr = [
join(args.dir_datasets, args.data_test + '/LR/X' + str(scale))
for scale in self.upscale
]
self.images_hr = [
entry.path for entry in scandir(self.dir_hr)
if is_image_file(entry.name)
]
self.images_lr = self._get_lr()
def __init__(self, opt):
base_dir = opt.train_dir
#self.use_npy = opt.use_npy
#if self.use_npy:
# low_dir = os.path.join(base_dir, 'low.npy')
# high_dir = os.path.join(base_dir, 'low_avg.npy')
low_dir = os.path.join(base_dir, 'Low')
high_dir = os.path.join(base_dir, 'Low_avg')
self.dsets = {}
#if self.use_npy:
# self.dsets['low'] = np.load(low_dir)
# self.dsets['high'] = np.load(high_dir)
self.dsets['low'] = [
os.path.join(high_dir, x) for x in os.listdir(low_dir)
if is_image_file(x)
]
self.dsets['high'] = [
os.path.join(high_dir, x) for x in os.listdir(high_dir)
if is_image_file(x)
]
def __init__(self, args):
super().__init__()
self.args = args
self.dir_demo = join(args.dir_datasets, args.data_test)
self.images_demo = [entry.path for entry in scandir(self.dir_demo) if is_image_file(entry.name)]