def TSS(root,
source_image_transform=None,
target_image_transform=None,
flow_transform=None,
co_transform=None,
split=None):
train_list, test_list = make_dataset(root, split)
train_dataset = ListDataset(root,
train_list,
source_image_transform=source_image_transform,
target_image_transform=target_image_transform,
flow_transform=flow_transform,
co_transform=co_transform,
loader=TSS_flow_loader,
mask=True,
size=True)
test_dataset = ListDataset(root,
test_list,
source_image_transform=source_image_transform,
target_image_transform=target_image_transform,
flow_transform=flow_transform,
co_transform=co_transform,
loader=TSS_flow_loader,
mask=True,
size=True)
return train_dataset, test_dataset
def KITTI_noc(root,
source_image_transform=None,
target_image_transform=None,
flow_transform=None,
co_transform=None,
split=None):
train_list, test_list = make_dataset(root, split, False)
train_dataset = ListDataset(root,
train_list,
source_image_transform=source_image_transform,
target_image_transform=target_image_transform,
flow_transform=flow_transform,
co_transform=co_transform,
loader=KITTI_flow_loader,
mask=True)
test_dataset = ListDataset(root,
test_list,
source_image_transform=source_image_transform,
target_image_transform=target_image_transform,
flow_transform=flow_transform,
co_transform=co_transform,
loader=KITTI_flow_loader,
mask=True)
return train_dataset, test_dataset
def kitti_occ_both(root,
source_image_transform=None,
target_image_transform=None,
flow_transform=None,
co_transform=None,
test_image_transform=None,
split=None):
train_list1, test_list1 = make_dataset(os.path.join(
root, 'KITTI_2012/training/'),
dataset_name='KITTI_2012/training/',
split=split,
occ=True)
train_list2, test_list2 = make_dataset(os.path.join(
root, 'KITTI_2015/training/'),
dataset_name='KITTI_2015/training/',
split=split,
occ=True)
train_dataset = ListDataset(root,
train_list1 + train_list2,
source_image_transform=source_image_transform,
target_image_transform=target_image_transform,
flow_transform=flow_transform,
co_transform=co_transform,
loader=KITTI_flow_loader,
mask=True)
if test_image_transform is None:
test_dataset = ListDataset(
root,
test_list1 + test_list2,
source_image_transform=source_image_transform,
target_image_transform=target_image_transform,
flow_transform=flow_transform,
co_transform=co_flow_and_images_transforms.CenterCrop((368, 1224)),
loader=KITTI_flow_loader,
mask=True)
else:
test_dataset = ListDataset(
root,
test_list1 + test_list2,
source_image_transform=test_image_transform,
target_image_transform=test_image_transform,
flow_transform=flow_transform,
co_transform=co_flow_and_images_transforms.CenterCrop((368, 1224)),
loader=KITTI_flow_loader,
mask=True)
return train_dataset, test_dataset
def fusion_data(root, transform=None, target_transform=None, co_transform=None, split=None):
train_list, test_list = make_dataset(root, split)
train_dataset = ListDataset(root, train_list, transform, target_transform, co_transform)
test_dataset = ListDataset(root, test_list, transform, target_transform, None)
return train_dataset, test_dataset