def get_loader(config):
# set the data loader
train_transforms = transforms.Compose([
d_utils.PointcloudToTensor(),
d_utils.PointcloudRandomRotate(x_range=config.x_angle_range,
y_range=config.y_angle_range,
z_range=config.z_angle_range),
d_utils.PointcloudScaleAndJitter(
scale_low=config.scale_low,
scale_high=config.scale_high,
std=config.noise_std,
clip=config.noise_clip,
augment_symmetries=config.augment_symmetries),
])
test_transforms = transforms.Compose([
d_utils.PointcloudToTensor(),
])
train_dataset = S3DISSeg(input_features_dim=config.input_features_dim,
subsampling_parameter=config.sampleDl,
color_drop=config.color_drop,
in_radius=config.in_radius,
num_points=config.num_points,
num_steps=config.num_steps,
num_epochs=config.epochs,
data_root=config.data_root,
transforms=train_transforms,
split='train')
val_dataset = S3DISSeg(input_features_dim=config.input_features_dim,
subsampling_parameter=config.sampleDl,
color_drop=config.color_drop,
in_radius=config.in_radius,
num_points=config.num_points,
num_steps=config.num_steps,
num_epochs=20,
data_root=config.data_root,
transforms=test_transforms,
split='val')
train_sampler = torch.utils.data.distributed.DistributedSampler(
train_dataset, shuffle=False)
train_loader = torch.utils.data.DataLoader(train_dataset,
batch_size=config.batch_size,
shuffle=False,
num_workers=config.num_workers,
pin_memory=True,
sampler=train_sampler,
drop_last=True)
val_sampler = torch.utils.data.distributed.DistributedSampler(
val_dataset, shuffle=False)
val_loader = torch.utils.data.DataLoader(val_dataset,
batch_size=config.batch_size,
shuffle=False,
num_workers=config.num_workers,
pin_memory=True,
sampler=val_sampler,
drop_last=False)
return train_loader, val_loader
def get_loader(args):
# set the data loader
train_transforms = transforms.Compose([
d_utils.PointcloudToTensor(),
d_utils.PointcloudScaleAndJitter(scale_low=config.scale_low,
scale_high=config.scale_high,
std=config.noise_std,
clip=config.noise_clip),
])
test_transforms = transforms.Compose([d_utils.PointcloudToTensor()])
train_dataset = PartNetSeg(input_features_dim=config.input_features_dim,
data_root=args.data_root,
transforms=train_transforms,
split='train')
val_dataset = PartNetSeg(input_features_dim=config.input_features_dim,
data_root=args.data_root,
transforms=test_transforms,
split='val')
test_dataset = PartNetSeg(input_features_dim=config.input_features_dim,
data_root=args.data_root,
transforms=test_transforms,
split='test')
train_sampler = torch.utils.data.distributed.DistributedSampler(
train_dataset)
train_loader = torch.utils.data.DataLoader(train_dataset,
batch_size=args.batch_size,
shuffle=False,
num_workers=args.num_workers,
pin_memory=True,
sampler=train_sampler,
drop_last=True)
val_sampler = torch.utils.data.distributed.DistributedSampler(
val_dataset, shuffle=False)
test_sampler = torch.utils.data.distributed.DistributedSampler(
test_dataset, shuffle=False)
val_loader = torch.utils.data.DataLoader(val_dataset,
batch_size=args.batch_size,
shuffle=False,
num_workers=args.num_workers,
pin_memory=True,
sampler=val_sampler,
drop_last=False)
test_loader = torch.utils.data.DataLoader(test_dataset,
batch_size=args.batch_size,
shuffle=False,
num_workers=args.num_workers,
pin_memory=True,
sampler=test_sampler,
drop_last=False)
return train_loader, val_loader, test_loader
def get_loader(args):
# set the data loader
test_transforms = transforms.Compose([d_utils.PointcloudToTensor()])
val_dataset = PartNetSeg(input_features_dim=config.input_features_dim,
data_root=args.data_root,
transforms=test_transforms,
split='val')
test_dataset = PartNetSeg(input_features_dim=config.input_features_dim,
data_root=args.data_root,
transforms=test_transforms,
split='test')
val_sampler = torch.utils.data.distributed.DistributedSampler(
val_dataset, shuffle=False)
test_sampler = torch.utils.data.distributed.DistributedSampler(
test_dataset, shuffle=False)
val_loader = torch.utils.data.DataLoader(val_dataset,
batch_size=args.batch_size,
shuffle=False,
num_workers=args.num_workers,
pin_memory=True,
sampler=val_sampler,
drop_last=False)
test_loader = torch.utils.data.DataLoader(test_dataset,
batch_size=args.batch_size,
shuffle=False,
num_workers=args.num_workers,
pin_memory=True,
sampler=test_sampler,
drop_last=False)
return val_loader, test_loader
def get_loader(args):
test_transforms = transforms.Compose([
d_utils.PointcloudToTensor()
])
test_dataset = ModelNet40Cls(input_features_dim=config.input_features_dim, num_points=args.num_points,
data_root=args.data_root, transforms=test_transforms,
subsampling_parameter=config.sampleDl,
split='test')
test_sampler = torch.utils.data.distributed.DistributedSampler(test_dataset, shuffle=False)
test_loader = torch.utils.data.DataLoader(test_dataset,
batch_size=args.batch_size,
shuffle=False,
num_workers=args.num_workers,
pin_memory=True,
sampler=test_sampler,
drop_last=False)
return test_loader
def get_loader(config):
test_transforms = transforms.Compose([
d_utils.PointcloudToTensor()
])
val_dataset = S3DISSemSeg(input_features_dim=config.input_features_dim,
subsampling_parameter=config.sampleDl, color_drop=config.color_drop,
in_radius=config.in_radius, num_points=config.num_points,
num_steps=config.num_steps, num_epochs=20,
transforms=test_transforms, split='val')
val_sampler = torch.utils.data.distributed.DistributedSampler(val_dataset, shuffle=False)
val_loader = torch.utils.data.DataLoader(val_dataset,
batch_size=config.batch_size,
shuffle=False,
num_workers=config.num_workers,
pin_memory=True,
sampler=val_sampler,
drop_last=False)
return val_loader
def get_loader(args):
# set the data loader
test_transforms = transforms.Compose([d_utils.PointcloudToTensor()])
test_dataset = ShapeNetPartSeg(num_points=args.num_points,
data_root=args.data_root,
transforms=test_transforms,
split='test')
test_sampler = torch.utils.data.distributed.DistributedSampler(
test_dataset, shuffle=False)
test_loader = torch.utils.data.DataLoader(test_dataset,
batch_size=args.batch_size,
shuffle=False,
num_workers=args.num_workers,
pin_memory=True,
sampler=test_sampler,
drop_last=False)
return test_loader
