def backbone_generator(params):
"""Generator function for various backbone models."""
if params.architecture.backbone == 'resnet':
resnet_params = params.resnet
backbone_fn = resnet.Resnet(
resnet_depth=resnet_params.resnet_depth,
dropblock=dropblock_generator(params.dropblock),
activation=params.batch_norm_activation.activation,
batch_norm_activation=batch_norm_activation_generator(
params.batch_norm_activation),
init_drop_connect_rate=resnet_params.init_drop_connect_rate)
elif params.architecture.backbone == 'spinenet':
spinenet_params = params.spinenet
block_specs_list = None
if spinenet_params.block_specs:
block_specs_list = json.loads(spinenet_params.block_specs)
backbone_fn = spinenet.spinenet_builder(
model_id=spinenet_params.model_id,
min_level=params.architecture.min_level,
max_level=params.architecture.max_level,
block_specs=spinenet.build_block_specs(block_specs_list),
use_native_resize_op=spinenet_params.use_native_resize_op,
activation=params.batch_norm_activation.activation,
batch_norm_activation=batch_norm_activation_generator(
params.batch_norm_activation),
init_drop_connect_rate=spinenet_params.init_drop_connect_rate)
elif params.architecture.backbone == 'spinenet_mbconv':
spinenet_mbconv_params = params.spinenet_mbconv
block_specs_list = None
if spinenet_mbconv_params.block_specs:
block_specs_list = json.loads(spinenet_mbconv_params.block_specs)
backbone_fn = spinenet_mbconv.spinenet_mbconv_builder(
model_id=spinenet_mbconv_params.model_id,
min_level=params.architecture.min_level,
max_level=params.architecture.max_level,
block_specs=spinenet_mbconv.build_block_specs(block_specs_list),
use_native_resize_op=spinenet_mbconv_params.use_native_resize_op,
se_ratio=spinenet_mbconv_params.se_ratio,
activation=params.batch_norm_activation.activation,
batch_norm_activation=batch_norm_activation_generator(
params.batch_norm_activation),
init_drop_connect_rate=spinenet_mbconv_params.
init_drop_connect_rate)
else:
raise ValueError('Backbone model %s is not supported.' %
params.architecture.backbone)
return backbone_fn
def backbone_generator(params):
"""Generator function for various backbone models."""
if params.architecture.backbone == 'resnet':
resnet_params = params.resnet
backbone_fn = resnet.Resnet(
resnet_depth=resnet_params.resnet_depth,
dropblock=dropblock_generator(params.dropblock),
activation=params.batch_norm_activation.activation,
batch_norm_activation=batch_norm_activation_generator(
params.batch_norm_activation),
init_drop_connect_rate=resnet_params.init_drop_connect_rate,
space_to_depth_block_size=params.architecture.
space_to_depth_block_size)
elif params.architecture.backbone == 'spinenet':
spinenet_params = params.spinenet
backbone_fn = spinenet.spinenet_builder(
model_id=spinenet_params.model_id,
min_level=params.architecture.min_level,
max_level=params.architecture.max_level,
use_native_resize_op=spinenet_params.use_native_resize_op,
activation=params.batch_norm_activation.activation,
batch_norm_activation=batch_norm_activation_generator(
params.batch_norm_activation),
init_drop_connect_rate=spinenet_params.init_drop_connect_rate)
elif params.architecture.backbone == 'spinenet_mbconv':
spinenet_mbconv_params = params.spinenet_mbconv
backbone_fn = spinenet_mbconv.spinenet_mbconv_builder(
model_id=spinenet_mbconv_params.model_id,
min_level=params.architecture.min_level,
max_level=params.architecture.max_level,
use_native_resize_op=spinenet_mbconv_params.use_native_resize_op,
se_ratio=spinenet_mbconv_params.se_ratio,
activation=params.batch_norm_activation.activation,
batch_norm_activation=batch_norm_activation_generator(
params.batch_norm_activation),
init_drop_connect_rate=spinenet_mbconv_params.
init_drop_connect_rate)
elif 'efficientnet' in params.architecture.backbone:
backbone_fn = efficientnet.Efficientnet(params.architecture.backbone)
else:
raise ValueError('Backbone model %s is not supported.' %
params.architecture.backbone)
return backbone_fn