import sys
sys.path.append('..')
from registry import Registry
META_ARCH_REGISTRY = Registry("META_ARCH")
META_ARCH_REGISTRY.__doc__ = """
Registry for meta-architectures, i.e. the whole model.
The registered object will be called with `obj(cfg)`
and expected to return a `nn.Module` object.
"""
__all__ = [
'build_model',
'META_ARCH_REGISTRY',
]
def build_model(cfg):
"""
Built the whole model, defined by `cfg.MODEL.META_ARCH`.
"""
meta_arch = cfg.model.name
return META_ARCH_REGISTRY.get(meta_arch)(cfg)
#coding=utf-8
from torchvision import transforms
from . import autoaugment
from registry import Registry
from utils import MyLogger
TRANSFORMS_REGISTRY = Registry('TRANSFORMS')
TRANSFORMS_REGISTRY.__doc__ = """
Registry for data transform functions, i.e. torchvision.transforms
The registered object will be called with `obj(cfg)`
"""
LABEL_TRANSFORMS_REGISTRY = Registry('LABEL_TRANSFORMS')
LABEL_TRANSFORMS_REGISTRY.__doc__ = """
Registry for label transform functions, i.e. torchvision.transforms
The registered object will be called with `obj(cfg)`
"""
__all__ = [
'build_transforms', 'build_label_transforms', 'TRANSFORMS_REGISTRY',
'LABEL_TRANSFORMS_REGISTRY', 'DefaultTransforms', 'BaseTransforms'
]
def build_transforms(cfg):
"""
Built the transforms, defined by `cfg.transforms.name`.
"""
import sys
sys.path.append('..')
from registry import Registry
LOSS_FN_REGISTRY = Registry("LOSS_FN")
LOSS_FN_REGISTRY.__doc__ = """
Registry for loss function, e.g. cross entropy loss.
The registered object will be called with `obj(cfg)`
"""
__all__ = ['build_loss_fn', 'LOSS_FN_REGISTRY']
def build_loss_fn(cfg):
"""
Built the loss function, defined by `cfg.loss.name`.
"""
name = cfg.loss.name
return LOSS_FN_REGISTRY.get(name)(cfg)
import sys
sys.path.append('..')
from registry import Registry
TRAINER_REGISTRY = Registry("TRAINER")
TRAINER_REGISTRY.__doc__ = """
Registry for trainer, i.e. the OnehotTrainer.
The registered object will be called with `obj(cfg)`
and expected to return a `nn.Module` object.
"""
__all__ = [
'build_trainer',
'TRAINER_REGISTRY',
]
def build_trainer(cfg):
"""
Built the trainer, defined by `cfg.trainer.name`.
"""
trainer = cfg.trainer.name
return TRAINER_REGISTRY.get(trainer)(cfg)
import sys
sys.path.append('..')
from registry import Registry
EVALUATOR_REGISTRY = Registry("TRAINER")
EVALUATOR_REGISTRY.__doc__ = """
Registry for evaluator, i.e. the DefaultEvaluator.
The registered object will be called with `obj(cfg)`
"""
__all__ = [
'build_evaluator',
'EVALUATOR_REGISTRY',
]
def build_evaluator(cfg):
"""
Built the trainer, defined by `cfg.trainer.name`.
"""
evaluator = cfg.evaluator.name
return EVALUATOR_REGISTRY.get(evaluator)(cfg)
import sys
sys.path.append('..')
from registry import Registry
MUTATOR_REGISTRY = Registry("MUTATOR")
MUTATOR_REGISTRY.__doc__ = """
Registry for mutator.
The registered object will be called with `obj(cfg)`
"""
__all__ = [
'MUTATOR_REGISTRY',
'build_mutator',
]
def build_mutator(model, cfg):
"""
Built the mutator.
"""
name = cfg.mutator.name
return MUTATOR_REGISTRY.get(name)(model, cfg)
import sys
sys.path.append('..')
from registry import Registry
DATASET_REGISTRY = Registry("DATASET")
DATASET_REGISTRY.__doc__ = """
Registry for dataset, i.e. torch.utils.data.Dataset.
The registered object will be called with `obj(cfg)`
"""
__all__ = [
'build_dataset',
'DATASET_REGISTRY'
]
def build_dataset(cfg):
"""
Built the dataset, defined by `cfg.dataset.name`.
"""
name = cfg.dataset.name
return DATASET_REGISTRY.get(name)(cfg)
