import torch
from torch import Tensor
import torch.nn as nn
from typing import Type, Any, Callable, Union, List, Optional
from backbone import MammothBackbone, register_backbone
__all__ = ['ResNet', 'resnet18', 'resnet34', 'resnet50', 'resnet101',
'resnet152', 'resnext50_32x4d', 'resnext101_32x8d',
'wide_resnet50_2', 'wide_resnet101_2']
model_urls = {
'resnet18': 'https://download.pytorch.org/models/resnet18-f37072fd.pth',
'resnet34': 'https://download.pytorch.org/models/resnet34-b627a593.pth',
'resnet50': 'https://download.pytorch.org/models/resnet50-0676ba61.pth',
'resnet101': 'https://download.pytorch.org/models/resnet101-63fe2227.pth',
'resnet152': 'https://download.pytorch.org/models/resnet152-394f9c45.pth',
'resnext50_32x4d': 'https://download.pytorch.org/models/resnext50_32x4d-7cdf4587.pth',
'resnext101_32x8d': 'https://download.pytorch.org/models/resnext101_32x8d-8ba56ff5.pth',
'wide_resnet50_2': 'https://download.pytorch.org/models/wide_resnet50_2-95faca4d.pth',
'wide_resnet101_2': 'https://download.pytorch.org/models/wide_resnet101_2-32ee1156.pth',
}
[docs]
def conv3x3(in_planes: int, out_planes: int, stride: int = 1, groups: int = 1, dilation: int = 1) -> nn.Conv2d:
"""3x3 convolution with padding"""
return nn.Conv2d(in_planes, out_planes, kernel_size=3, stride=stride,
padding=dilation, groups=groups, bias=False, dilation=dilation)
[docs]
def conv1x1(in_planes: int, out_planes: int, stride: int = 1) -> nn.Conv2d:
"""1x1 convolution"""
return nn.Conv2d(in_planes, out_planes, kernel_size=1, stride=stride, bias=False)
[docs]
class Bottleneck(nn.Module):
# Bottleneck in torchvision places the stride for downsampling at 3x3 convolution(self.conv2)
# while original implementation places the stride at the first 1x1 convolution(self.conv1)
# according to "Deep residual learning for image recognition"https://arxiv.org/abs/1512.03385.
# This variant is also known as ResNet V1.5 and improves accuracy according to
# https://ngc.nvidia.com/catalog/model-scripts/nvidia:resnet_50_v1_5_for_pytorch.
expansion: int = 4
def __init__(
self,
inplanes: int,
planes: int,
stride: int = 1,
downsample: Optional[nn.Module] = None,
groups: int = 1,
base_width: int = 64,
dilation: int = 1,
norm_layer: Optional[Callable[..., nn.Module]] = None
) -> None:
super(Bottleneck, self).__init__()
self.return_prerelu = False
if norm_layer is None:
norm_layer = nn.BatchNorm2d
width = int(planes * (base_width / 64.)) * groups
# Both self.conv2 and self.downsample layers downsample the input when stride != 1
self.conv1 = conv1x1(inplanes, width)
self.bn1 = norm_layer(width)
self.conv2 = conv3x3(width, width, stride, groups, dilation)
self.bn2 = norm_layer(width)
self.conv3 = conv1x1(width, planes * self.expansion)
self.bn3 = norm_layer(planes * self.expansion)
self.relu = nn.ReLU(inplace=True)
self.downsample = downsample
self.stride = stride
[docs]
def forward(self, x: Tensor) -> Tensor:
identity = x
out = self.conv1(x)
out = self.bn1(out)
out = self.relu(out)
out = self.conv2(out)
out = self.bn2(out)
out = self.relu(out)
out = self.conv3(out)
out = self.bn3(out)
if self.downsample is not None:
identity = self.downsample(x)
out += identity
if self.return_prerelu:
self.prerelu = out.clone()
out = self.relu(out)
return out
[docs]
class ResNet(MammothBackbone):
def __init__(
self,
block: Bottleneck,
layers: List[int],
num_classes: int = 1000,
zero_init_residual: bool = False,
groups: int = 1,
pretrained: bool = False,
width_per_group: int = 64,
replace_stride_with_dilation: Optional[List[bool]] = None,
norm_layer: Optional[Callable[..., nn.Module]] = None
) -> None:
super(ResNet, self).__init__()
self.block = block
self.device = "cpu"
if norm_layer is None:
norm_layer = nn.BatchNorm2d
self._norm_layer = norm_layer
self.return_prerelu = False
self.num_classes = num_classes
self.inplanes = 64
self.dilation = 1
if replace_stride_with_dilation is None:
# each element in the tuple indicates if we should replace
# the 2x2 stride with a dilated convolution instead
replace_stride_with_dilation = [False, False, False]
if len(replace_stride_with_dilation) != 3:
raise ValueError("replace_stride_with_dilation should be None "
"or a 3-element tuple, got {}".format(replace_stride_with_dilation))
self.groups = groups
self.base_width = width_per_group
self.conv1 = nn.Conv2d(3, self.inplanes, kernel_size=7, stride=2, padding=3,
bias=False)
self.bn1 = norm_layer(self.inplanes)
self.relu = nn.ReLU(inplace=True)
self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)
self.layer1 = self._make_layer(block, 64, layers[0])
self.layer2 = self._make_layer(block, 128, layers[1], stride=2,
dilate=replace_stride_with_dilation[0])
self.layer3 = self._make_layer(block, 256, layers[2], stride=2,
dilate=replace_stride_with_dilation[1])
self.layer4 = self._make_layer(block, 512, layers[3], stride=2,
dilate=replace_stride_with_dilation[2])
self.avgpool = nn.AdaptiveAvgPool2d((1, 1))
self.classifier = nn.Linear(512 * block.expansion, num_classes)
self.feature_dim = 512 * block.expansion
for m in self.modules():
if isinstance(m, nn.Conv2d):
nn.init.kaiming_normal_(
m.weight, mode='fan_out', nonlinearity='relu')
elif isinstance(m, (nn.BatchNorm2d, nn.GroupNorm)):
nn.init.constant_(m.weight, 1)
nn.init.constant_(m.bias, 0)
# Zero-initialize the last BN in each residual branch,
# so that the residual branch starts with zeros, and each residual block behaves like an identity.
# This improves the model by 0.2~0.3% according to https://arxiv.org/abs/1706.02677
if zero_init_residual:
for m in self.modules():
if isinstance(m, Bottleneck):
# type: ignore[arg-type]
nn.init.constant_(m.bn3.weight, 0)
if pretrained:
ckpt = torch.hub.load_state_dict_from_url(model_urls['resnet50'], progress=True, check_hash=True)
# drop classifier weights
ckpt.pop('fc.weight')
ckpt.pop('fc.bias')
missing, unexpected = self.load_state_dict(ckpt, strict=False)
assert len([x for x in missing if 'classifier' not in x]) == 0, "Missing keys: {}".format(missing)
assert len(unexpected) == 0, "Unexpected keys: {}".format(unexpected)
[docs]
def to(self, device, **kwargs):
self.device = device
return super().to(device, **kwargs)
[docs]
def set_return_prerelu(self, enable=True):
self.return_prerelu = enable
for c in self.modules():
if isinstance(c, self.block):
c.return_prerelu = enable
def _make_layer(self, block: Bottleneck, planes: int, blocks: int,
stride: int = 1, dilate: bool = False) -> nn.Sequential:
norm_layer = self._norm_layer
downsample = None
previous_dilation = self.dilation
if dilate:
self.dilation *= stride
stride = 1
if stride != 1 or self.inplanes != planes * block.expansion:
downsample = nn.Sequential(
conv1x1(self.inplanes, planes * block.expansion, stride),
norm_layer(planes * block.expansion),
)
layers = []
layers.append(block(self.inplanes, planes, stride, downsample, self.groups,
self.base_width, previous_dilation, norm_layer))
self.inplanes = planes * block.expansion
for _ in range(1, blocks):
layers.append(block(self.inplanes, planes, groups=self.groups,
base_width=self.base_width, dilation=self.dilation,
norm_layer=norm_layer))
return nn.Sequential(*layers)
[docs]
def forward(self, x: Tensor, returnt="out") -> Tensor:
out_0 = self.conv1(x)
out_0 = self.bn1(out_0)
if self.return_prerelu:
out_0_t = out_0.clone()
out_0 = self.relu(out_0)
out_0 = self.maxpool(out_0)
out_1 = self.layer1(out_0)
out_2 = self.layer2(out_1)
out_3 = self.layer3(out_2)
out_4 = self.layer4(out_3)
feature = self.avgpool(out_4)
feature = torch.flatten(feature, 1)
if returnt == 'features':
return feature
out = self.classifier(feature)
if returnt == 'out':
return out
elif returnt == 'full':
return out, [
out_0 if not self.return_prerelu else out_0_t,
out_1 if not self.return_prerelu else self.layer1[-1].prerelu,
out_2 if not self.return_prerelu else self.layer2[-1].prerelu,
out_3 if not self.return_prerelu else self.layer3[-1].prerelu,
out_4 if not self.return_prerelu else self.layer4[-1].prerelu
]
elif returnt == 'both':
return (out, feature)
raise NotImplementedError("Unknown return type. Must be in ['out', 'features', 'both', 'full'] but got {}".format(returnt))
[docs]
def set_grad_filter(self, filter_s: str, enable: bool) -> None:
negative_mode = filter_s[0] == '~'
if negative_mode:
filter_s = filter_s[1:]
for _, p in filter(lambda x: filter_s not in x[0], self.named_parameters()):
p.requires_grad = enable
else:
for _, p in filter(lambda x: filter_s in x[0], self.named_parameters()):
p.requires_grad = enable
@register_backbone("resnet50")
def resnet50(num_classes: int, pretrained=False) -> ResNet:
r"""ResNet-50 model from
`"Deep Residual Learning for Image Recognition" <https://arxiv.org/pdf/1512.03385.pdf>`_.
Args:
pretrained (bool): If True, returns a model pre-trained on ImageNet
progress (bool): If True, displays a progress bar of the download to stderr
"""
return ResNet(Bottleneck, [3, 4, 6, 3], num_classes=num_classes, pretrained=pretrained)
[docs]
@register_backbone("resnet50_pt")
def resnet50(num_classes: int) -> ResNet:
r"""ResNet-50 model from
`"Deep Residual Learning for Image Recognition" <https://arxiv.org/pdf/1512.03385.pdf>`_.
Args:
pretrained (bool): If True, returns a model pre-trained on ImageNet
progress (bool): If True, displays a progress bar of the download to stderr
"""
return ResNet(Bottleneck, [3, 4, 6, 3], num_classes=num_classes, pretrained=True)