29 lines
878 B
Python
29 lines
878 B
Python
# dataset settings
|
|
dataset_type = 'mmcls.ImageNet'
|
|
data_root = 'data/imagenet/'
|
|
file_client_args = dict(backend='disk')
|
|
|
|
train_pipeline = [
|
|
dict(type='LoadImageFromFile', file_client_args=file_client_args),
|
|
dict(type='Resize', scale=292),
|
|
dict(type='RandomCrop', size=255),
|
|
dict(type='RandomGrayscale', prob=0.66, keep_channels=True),
|
|
dict(type='RandomPatchWithLabels'),
|
|
dict(
|
|
type='PackSelfSupInputs',
|
|
pseudo_label_keys=['patch_box', 'patch_label', 'unpatched_img'],
|
|
meta_keys=['img_path'])
|
|
]
|
|
|
|
train_dataloader = dict(
|
|
batch_size=64,
|
|
num_workers=4,
|
|
persistent_workers=True,
|
|
sampler=dict(type='DefaultSampler', shuffle=True),
|
|
dataset=dict(
|
|
type=dataset_type,
|
|
data_root=data_root,
|
|
ann_file='meta/train.txt',
|
|
data_prefix=dict(img_path='train/'),
|
|
pipeline=train_pipeline))
|