"""
Modified from https://github.com/pytorch/vision/blob/main/torchvision/models/vgg.py
"""
from typing import Union, List, Dict, Any, cast
import oneflow as flow
import oneflow.nn as nn
from .utils import load_state_dict_from_url
from .registry import ModelCreator
__all__ = [
"VGG",
"vgg11",
"vgg11_bn",
"vgg13",
"vgg13_bn",
"vgg16",
"vgg16_bn",
"vgg19",
"vgg19_bn",
]
model_urls = {
"vgg11": "https://oneflow-public.oss-cn-beijing.aliyuncs.com/model_zoo/flowvision/classification/VGG/vgg11.zip",
"vgg13": "https://oneflow-public.oss-cn-beijing.aliyuncs.com/model_zoo/flowvision/classification/VGG/vgg13.zip",
"vgg16": "https://oneflow-public.oss-cn-beijing.aliyuncs.com/model_zoo/flowvision/classification/VGG/vgg16.zip",
"vgg19": "https://oneflow-public.oss-cn-beijing.aliyuncs.com/model_zoo/flowvision/classification/VGG/vgg19.zip",
"vgg11_bn": "https://oneflow-public.oss-cn-beijing.aliyuncs.com/model_zoo/flowvision/classification/VGG/vgg11_bn.zip",
"vgg13_bn": "https://oneflow-public.oss-cn-beijing.aliyuncs.com/model_zoo/flowvision/classification/VGG/vgg13_bn.zip",
"vgg16_bn": "https://oneflow-public.oss-cn-beijing.aliyuncs.com/model_zoo/flowvision/classification/VGG/vgg16_bn.zip",
"vgg19_bn": "https://oneflow-public.oss-cn-beijing.aliyuncs.com/model_zoo/flowvision/classification/VGG/vgg19_bn.zip",
}
class VGG(nn.Module):
def __init__(
self, features: nn.Module, num_classes: int = 1000, init_weights: bool = True
) -> None:
super(VGG, self).__init__()
self.features = features
self.avgpool = nn.AdaptiveAvgPool2d((7, 7))
self.classifier = nn.Sequential(
nn.Linear(512 * 7 * 7, 4096),
nn.ReLU(True),
nn.Dropout(),
nn.Linear(4096, 4096),
nn.ReLU(True),
nn.Dropout(),
nn.Linear(4096, num_classes),
)
if init_weights:
self._initialize_weights()
def forward(self, x: flow.Tensor) -> flow.Tensor:
x = self.features(x)
x = self.avgpool(x)
x = flow.flatten(x, 1)
x = self.classifier(x)
return x
def _initialize_weights(self) -> None:
for m in self.modules():
if isinstance(m, nn.Conv2d):
nn.init.kaiming_normal_(m.weight, mode="fan_out", nonlinearity="relu")
if m.bias is not None:
nn.init.constant_(m.bias, 0)
elif isinstance(m, nn.BatchNorm2d):
nn.init.constant_(m.weight, 1)
nn.init.constant_(m.bias, 0)
elif isinstance(m, nn.Linear):
nn.init.normal_(m.weight, 0, 0.01)
nn.init.constant_(m.bias, 0)
def make_layers(cfg: List[Union[str, int]], batch_norm: bool = False) -> nn.Sequential:
layers: List[nn.Module] = []
in_channels = 3
for v in cfg:
if v == "M":
layers += [nn.MaxPool2d(kernel_size=2, stride=2)]
else:
v = cast(int, v)
conv2d = nn.Conv2d(in_channels, v, kernel_size=3, padding=1)
if batch_norm:
layers += [conv2d, nn.BatchNorm2d(v), nn.ReLU(inplace=True)]
else:
layers += [conv2d, nn.ReLU(inplace=True)]
in_channels = v
return nn.Sequential(*layers)
cfgs: Dict[str, List[Union[str, int]]] = {
"A": [64, "M", 128, "M", 256, 256, "M", 512, 512, "M", 512, 512, "M"],
"B": [64, 64, "M", 128, 128, "M", 256, 256, "M", 512, 512, "M", 512, 512, "M"],
"D": [
64,
64,
"M",
128,
128,
"M",
256,
256,
256,
"M",
512,
512,
512,
"M",
512,
512,
512,
"M",
],
"E": [
64,
64,
"M",
128,
128,
"M",
256,
256,
256,
256,
"M",
512,
512,
512,
512,
"M",
512,
512,
512,
512,
"M",
],
}
def _vgg(
arch: str,
cfg: str,
batch_norm: bool,
pretrained: bool,
progress: bool,
**kwargs: Any
) -> VGG:
if pretrained:
kwargs["init_weights"] = False
model = VGG(make_layers(cfgs[cfg], batch_norm=batch_norm), **kwargs)
if pretrained:
state_dict = load_state_dict_from_url(model_urls[arch], progress=progress)
model.load_state_dict(state_dict)
return model
[docs]@ModelCreator.register_model
def vgg11(pretrained: bool = False, progress: bool = True, **kwargs: Any) -> VGG:
"""
Constructs the VGG-11 model (configuration "A").
.. note::
VGG 11-layer model (configuration “A”) from `"Very Deep Convolutional Networks For Large-Scale Image Recognition" <https://arxiv.org/pdf/1409.1556.pdf>`_.
The required minimum input size of the model is 32x32.
Args:
pretrained (bool): Whether to download the pre-trained model on ImageNet. Default: ``False``
progress (bool): If True, displays a progress bar of the download to stderr. Default: ``True``
For example:
.. code-block:: python
>>> import flowvision
>>> vgg11 = flowvision.models.vgg11(pretrained=False, progress=True)
"""
return _vgg("vgg11", "A", False, pretrained, progress, **kwargs)
[docs]@ModelCreator.register_model
def vgg11_bn(pretrained: bool = False, progress: bool = True, **kwargs: Any) -> VGG:
"""
Constructs the VGG-11 model with batch normalization (configuration "A").
.. note::
VGG 11-layer model (configuration “A”) with batch normalization `"Very Deep Convolutional Networks For Large-Scale Image Recognition" <https://arxiv.org/pdf/1409.1556.pdf>`_.
The required minimum input size of the model is 32x32.
Args:
pretrained (bool): Whether to download the pre-trained model on ImageNet. Default: ``False``
progress (bool): If True, displays a progress bar of the download to stderr. Default: ``True``
For example:
.. code-block:: python
>>> import flowvision
>>> vgg11_bn = flowvision.models.vgg11_bn(pretrained=False, progress=True)
"""
return _vgg("vgg11_bn", "A", True, pretrained, progress, **kwargs)
[docs]@ModelCreator.register_model
def vgg13(pretrained: bool = False, progress: bool = True, **kwargs: Any) -> VGG:
"""
Constructs the VGG-13 model (configuration "B").
.. note::
VGG 13-layer model (configuration “B”) from `"Very Deep Convolutional Networks For Large-Scale Image Recognition" <https://arxiv.org/pdf/1409.1556.pdf>`_.
The required minimum input size of the model is 32x32.
Args:
pretrained (bool): Whether to download the pre-trained model on ImageNet. Default: ``False``
progress (bool): If True, displays a progress bar of the download to stderr. Default: ``True``
For example:
.. code-block:: python
>>> import flowvision
>>> vgg13 = flowvision.models.vgg13(pretrained=False, progress=True)
"""
return _vgg("vgg13", "B", False, pretrained, progress, **kwargs)
[docs]@ModelCreator.register_model
def vgg13_bn(pretrained: bool = False, progress: bool = True, **kwargs: Any) -> VGG:
"""
Constructs the VGG-13 model with batch normalization (configuration "B").
.. note::
VGG 13-layer model (configuration “B”) with batch normalization from `"Very Deep Convolutional Networks For Large-Scale Image Recognition" <https://arxiv.org/pdf/1409.1556.pdf>`_.
The required minimum input size of the model is 32x32.
Args:
pretrained (bool): Whether to download the pre-trained model on ImageNet. Default: ``False``
progress (bool): If True, displays a progress bar of the download to stderr. Default: ``True``
For example:
.. code-block:: python
>>> import flowvision
>>> vgg13_bn = flowvision.models.vgg13_bn(pretrained=False, progress=True)
"""
return _vgg("vgg13_bn", "B", True, pretrained, progress, **kwargs)
[docs]@ModelCreator.register_model
def vgg16(pretrained: bool = False, progress: bool = True, **kwargs: Any) -> VGG:
"""
Constructs the VGG-16 model (configuration "D").
.. note::
VGG 16-layer model (configuration “D”) from `"Very Deep Convolutional Networks For Large-Scale Image Recognition" <https://arxiv.org/pdf/1409.1556.pdf>`_.
The required minimum input size of the model is 32x32.
Args:
pretrained (bool): Whether to download the pre-trained model on ImageNet. Default: ``False``
progress (bool): If True, displays a progress bar of the download to stderr. Default: ``True``
For example:
.. code-block:: python
>>> import flowvision
>>> vgg16 = flowvision.models.vgg16(pretrained=False, progress=True)
"""
return _vgg("vgg16", "D", False, pretrained, progress, **kwargs)
[docs]@ModelCreator.register_model
def vgg16_bn(pretrained: bool = False, progress: bool = True, **kwargs: Any) -> VGG:
"""
Constructs the VGG-16 model (configuration "D") with batch normalization.
.. note::
VGG 16-layer model (configuration “D”) with batch normalization from `"Very Deep Convolutional Networks For Large-Scale Image Recognition" <https://arxiv.org/pdf/1409.1556.pdf>`_.
The required minimum input size of the model is 32x32.
Args:
pretrained (bool): Whether to download the pre-trained model on ImageNet. Default: ``False``
progress (bool): If True, displays a progress bar of the download to stderr. Default: ``True``
For example:
.. code-block:: python
>>> import flowvision
>>> vgg16_bn = flowvision.models.vgg16_bn(pretrained=False, progress=True)
"""
return _vgg("vgg16_bn", "D", True, pretrained, progress, **kwargs)
[docs]@ModelCreator.register_model
def vgg19(pretrained: bool = False, progress: bool = True, **kwargs: Any) -> VGG:
"""
Constructs the VGG-19 model (configuration "E").
.. note::
VGG 19-layer model (configuration “E”) from `"Very Deep Convolutional Networks For Large-Scale Image Recognition" <https://arxiv.org/pdf/1409.1556.pdf>`_.
The required minimum input size of the model is 32x32.
Args:
pretrained (bool): Whether to download the pre-trained model on ImageNet. Default: ``False``
progress (bool): If True, displays a progress bar of the download to stderr. Default: ``True``
For example:
.. code-block:: python
>>> import flowvision
>>> vgg19 = flowvision.models.vgg19(pretrained=False, progress=True)
"""
return _vgg("vgg19", "E", False, pretrained, progress, **kwargs)
[docs]@ModelCreator.register_model
def vgg19_bn(pretrained: bool = False, progress: bool = True, **kwargs: Any) -> VGG:
"""
Constructs the VGG-19 model (configuration "E") with batch normalization.
.. note::
VGG 19-layer model (configuration “E”) with batch normalization from `"Very Deep Convolutional Networks For Large-Scale Image Recognition" <https://arxiv.org/pdf/1409.1556.pdf>`_.
The required minimum input size of the model is 32x32.
Args:
pretrained (bool): Whether to download the pre-trained model on ImageNet. Default: ``False``
progress (bool): If True, displays a progress bar of the download to stderr. Default: ``True``
For example:
.. code-block:: python
>>> import flowvision
>>> vgg19_bn = flowvision.models.vgg19_bn(pretrained=False, progress=True)
"""
return _vgg("vgg19_bn", "E", True, pretrained, progress, **kwargs)