"""
Modified from https://github.com/Res2Net/Res2Net-PretrainedModels/blob/master/res2net.py
"""
import math
import oneflow as flow
import oneflow.nn as nn
import oneflow.nn.functional as F
from .utils import load_state_dict_from_url
from .registry import ModelCreator
model_urls = {
"res2net50_26w_4s": "https://oneflow-public.oss-cn-beijing.aliyuncs.com/model_zoo/flowvision/classification/Res2Net/res2net50_26w_4s.zip",
"res2net50_26w_6s": "https://oneflow-public.oss-cn-beijing.aliyuncs.com/model_zoo/flowvision/classification/Res2Net/res2net50_26w_6s.zip",
"res2net50_26w_8s": "https://oneflow-public.oss-cn-beijing.aliyuncs.com/model_zoo/flowvision/classification/Res2Net/res2net50_26w_8s.zip",
"res2net50_48w_2s": "https://oneflow-public.oss-cn-beijing.aliyuncs.com/model_zoo/flowvision/classification/Res2Net/res2net50_48w_2s.zip",
"res2net50_14w_8s": "https://oneflow-public.oss-cn-beijing.aliyuncs.com/model_zoo/flowvision/classification/Res2Net/res2net50_14w_8s.zip",
"res2net101_26w_4s": "https://oneflow-public.oss-cn-beijing.aliyuncs.com/model_zoo/flowvision/classification/Res2Net/res2net101_26w_4s.zip",
}
class Bottle2neck(nn.Module):
expansion = 4
def __init__(
self,
inplanes,
planes,
stride=1,
downsample=None,
baseWidth=26,
scale=4,
stype="normal",
):
""" Constructor
Args:
inplanes: Input channel dimensionality
planes: Output channel dimensionality
stride: Conv stride. Replaces pooling layer
downsample: None when stride = 1
baseWidth: Basic width of conv3x3
scale: Number of scale
stype: 'normal': normal set. 'stage': first block of a new stage
"""
super(Bottle2neck, self).__init__()
width = int(math.floor(planes * (baseWidth / 64.0)))
self.conv1 = nn.Conv2d(inplanes, width * scale, kernel_size=1, bias=False)
self.bn1 = nn.BatchNorm2d(width * scale)
if scale == 1:
self.nums = 1
else:
self.nums = scale - 1
if stype == "stage":
self.pool = nn.AvgPool2d(kernel_size=3, stride=stride, padding=1)
convs = []
bns = []
for i in range(self.nums):
convs.append(
nn.Conv2d(
width, width, kernel_size=3, stride=stride, padding=1, bias=False
)
)
bns.append(nn.BatchNorm2d(width))
self.convs = nn.ModuleList(convs)
self.bns = nn.ModuleList(bns)
self.conv3 = nn.Conv2d(
width * scale, planes * self.expansion, kernel_size=1, bias=False
)
self.bn3 = nn.BatchNorm2d(planes * self.expansion)
self.relu = nn.ReLU(inplace=True)
self.downsample = downsample
self.stype = stype
self.scale = scale
self.width = width
def forward(self, x):
residual = x
out = self.conv1(x)
out = self.bn1(out)
out = self.relu(out)
spx = flow.split(out, self.width, 1)
for i in range(self.nums):
if i == 0 or self.stype == "stage":
sp = spx[i]
else:
sp = sp + spx[i]
sp = self.convs[i](sp)
sp = self.relu(self.bns[i](sp))
if i == 0:
out = sp
else:
out = flow.cat((out, sp), dim=1)
if self.scale != 1 and self.stype == "normal":
out = flow.cat((out, spx[self.nums]), 1)
elif self.scale != 1 and self.stype == "stage":
out = flow.cat((out, self.pool(spx[self.nums])), 1)
out = self.conv3(out)
out = self.bn3(out)
if self.downsample is not None:
residual = self.downsample(x)
out += residual
out = self.relu(out)
return out
class Res2Net(nn.Module):
def __init__(self, block, layers, baseWidth=26, scale=4, num_classes=1000):
self.inplanes = 64
super(Res2Net, self).__init__()
self.baseWidth = baseWidth
self.scale = scale
self.conv1 = nn.Conv2d(3, 64, kernel_size=7, stride=2, padding=3, bias=False)
self.bn1 = nn.BatchNorm2d(64)
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)
self.layer3 = self._make_layer(block, 256, layers[2], stride=2)
self.layer4 = self._make_layer(block, 512, layers[3], stride=2)
self.avgpool = nn.AdaptiveAvgPool2d(1)
self.fc = nn.Linear(512 * block.expansion, num_classes)
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.init.constant_(m.weight, 1)
nn.init.constant_(m.bias, 0)
def _make_layer(self, block, planes, blocks, stride=1):
downsample = None
if stride != 1 or self.inplanes != planes * block.expansion:
downsample = nn.Sequential(
nn.Conv2d(
self.inplanes,
planes * block.expansion,
kernel_size=1,
stride=stride,
bias=False,
),
nn.BatchNorm2d(planes * block.expansion),
)
layers = []
layers.append(
block(
self.inplanes,
planes,
stride,
downsample=downsample,
stype="stage",
baseWidth=self.baseWidth,
scale=self.scale,
)
)
self.inplanes = planes * block.expansion
for i in range(1, blocks):
layers.append(
block(self.inplanes, planes, baseWidth=self.baseWidth, scale=self.scale)
)
return nn.Sequential(*layers)
def forward(self, x):
x = self.conv1(x)
x = self.bn1(x)
x = self.relu(x)
x = self.maxpool(x)
x = self.layer1(x)
x = self.layer2(x)
x = self.layer3(x)
x = self.layer4(x)
x = self.avgpool(x)
x = x.view(x.size(0), -1)
x = self.fc(x)
return x
def _create_res2net(arch, pretrained=False, progress=True, **model_kwargs):
model = Res2Net(**model_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 res2net50_26w_4s(pretrained=False, progress=True, **kwargs):
"""
Constructs the Res2Net-50_26w_4s model.
.. note::
Res2Net-50_26w_4s model from the `Res2Net: A New Multi-scale Backbone Architecture <https://arxiv.org/pdf/1904.01169.pdf>`_ paper.
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
>>> res2net50_26w_4s = flowvision.models.res2net50_26w_4s(pretrained=False, progress=True)
"""
model_kwargs = dict(
block=Bottle2neck, layers=[3, 4, 6, 3], baseWidth=26, scale=4, **kwargs
)
return _create_res2net(
"res2net50_26w_4s", pretrained=pretrained, progress=progress, **model_kwargs
)
[docs]@ModelCreator.register_model
def res2net101_26w_4s(pretrained=False, progress=True, **kwargs):
"""
Constructs the Res2Net-101_26w_4s model.
.. note::
Res2Net-101_26w_4s model from the `Res2Net: A New Multi-scale Backbone Architecture <https://arxiv.org/pdf/1904.01169.pdf>`_ paper.
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
>>> res2net101_26w_4s = flowvision.models.res2net101_26w_4s(pretrained=False, progress=True)
"""
model_kwargs = dict(
block=Bottle2neck, layers=[3, 4, 23, 3], baseWidth=26, scale=4, **kwargs
)
return _create_res2net(
"res2net101_26w_4s", pretrained=pretrained, progress=progress, **model_kwargs
)
[docs]@ModelCreator.register_model
def res2net50_26w_6s(pretrained=False, progress=True, **kwargs):
"""
Constructs the Res2Net-50_26w_6s model.
.. note::
Res2Net-50_26w_6s model from the `Res2Net: A New Multi-scale Backbone Architecture <https://arxiv.org/pdf/1904.01169.pdf>`_ paper.
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
>>> res2net50_26w_6s = flowvision.models.res2net50_26w_6s(pretrained=False, progress=True)
"""
model_kwargs = dict(
block=Bottle2neck, layers=[3, 4, 6, 3], baseWidth=26, scale=6, **kwargs
)
return _create_res2net(
"res2net50_26w_6s", pretrained=pretrained, progress=progress, **model_kwargs
)
[docs]@ModelCreator.register_model
def res2net50_26w_8s(pretrained=False, progress=True, **kwargs):
"""
Constructs the Res2Net-50_26w_8s model.
.. note::
Res2Net-50_26w_8s model from the `Res2Net: A New Multi-scale Backbone Architecture <https://arxiv.org/pdf/1904.01169.pdf>`_ paper.
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
>>> res2net50_26w_8s = flowvision.models.res2net50_26w_8s(pretrained=False, progress=True)
"""
model_kwargs = dict(
block=Bottle2neck, layers=[3, 4, 6, 3], baseWidth=26, scale=8, **kwargs
)
return _create_res2net(
"res2net50_26w_8s", pretrained=pretrained, progress=progress, **model_kwargs
)
[docs]@ModelCreator.register_model
def res2net50_48w_2s(pretrained=False, progress=True, **kwargs):
"""
Constructs the Res2Net-50_48w_2s model.
.. note::
Res2Net-50_48w_2s model from the `Res2Net: A New Multi-scale Backbone Architecture <https://arxiv.org/pdf/1904.01169.pdf>`_ paper.
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
>>> res2net50_48w_2s = flowvision.models.res2net50_48w_2s(pretrained=False, progress=True)
"""
model_kwargs = dict(
block=Bottle2neck, layers=[3, 4, 6, 3], baseWidth=48, scale=2, **kwargs
)
return _create_res2net(
"res2net50_48w_2s", pretrained=pretrained, progress=progress, **model_kwargs
)
[docs]@ModelCreator.register_model
def res2net50_14w_8s(pretrained=False, progress=True, **kwargs):
"""
Constructs the Res2Net-50_14w_8s model.
.. note::
Res2Net-50_14w_8s model from the `Res2Net: A New Multi-scale Backbone Architecture <https://arxiv.org/pdf/1904.01169.pdf>`_ paper.
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
>>> res2net50_14w_8s = flowvision.models.res2net50_14w_8s(pretrained=False, progress=True)
"""
model_kwargs = dict(
block=Bottle2neck, layers=[3, 4, 6, 3], baseWidth=14, scale=8, **kwargs
)
return _create_res2net(
"res2net50_14w_8s", pretrained=pretrained, progress=progress, **model_kwargs
)