ResNext(50和101)模型构建(可以使用torchvision预训练权重)
ResNeXt是一种卷积神经网络，它由Xie等人在论文《Aggregated Residual Transformations for Deep Neural Networks》中提出¹。ResNeXt结合了ResNet和Inception的优点，但不同于Inception v4，ResNeXt不需要人工设计复杂的Inception结构细节，而是每一个分支都采用相同的拓扑结构¹。
ResNeXt的本质是分组卷积（Group Convolution），通过变量基数（Cardinality）来控制组的数量¹。与ResNet相比，它增加了一个新维度——基数（一组转换的大小），作为深度和宽度之外的一个重要因素²。
(1) ResNeXt详解 - 知乎. https://zhuanlan.zhihu.com/p/51075096.
(2) ResNeXt Explained | Papers With Code. https://paperswithcode.com/method/resnext.
(3) ResNext | PyTorch. https://pytorch.org/hub/pytorch_vision_resnext/.
ResNext的代码主要参考了ResNet的构建，ResNet代码可以参考这一篇博客。
https://blog.csdn.net/qq_44733260/article/details/131340430
import math
import numpy as np
import torch.nn as nn
from torch.hub import load_state_dict_from_url
import torchvision.models as models
from torchsummary import summary
class Bottleneck(nn.Module):
expansion = 4
def __init__(self,inplanes,planes,stride=1,downsample=None,groups=1,base_width=4,base_channels=64):
super(Bottleneck,self).__init__()
if groups==1:
width = planes*2
else:
width = (math.floor(planes *(base_width / base_channels)) * groups)*2
self.conv1 = nn.Conv2d(inplanes, width, kernel_size=1, stride=stride,bias=False)
self.bn1 = nn.BatchNorm2d(width)
self.conv2 = nn.Conv2d(width, width, kernel_size=3, stride=1, padding=1,groups=32,bias=False)
self.bn2 = nn.BatchNorm2d(width)
self.conv3 = nn.Conv2d(width, planes * 4, kernel_size=1,stride=1,bias=False)
self.bn3 = nn.BatchNorm2d(planes * 4)
self.relu = nn.ReLU(inplace=True)
self.downsample = downsample
self.stride = stride
def forward(self, x):
residual = 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:
residual = self.downsample(x)
out += residual
out = self.relu(out)
return out
class Bottleneck_50(Bottleneck):
def __init__(self,inplanes,planes,stride=1,downsample=None,groups=1,base_width=4,base_channels=64):
super().__init__(inplanes,planes,stride,downsample,groups,base_width,base_channels)
class Bottleneck_101(Bottleneck):
def __init__(self,inplanes,planes,stride=1,downsample=None,groups=32,base_width=4,base_channels=64):
super().__init__(inplanes,planes,stride,downsample,groups,base_width,base_channels)
class ResNext(nn.Module):
def __init__(self, block, layers, num_classes=1000):
self.inplanes = 64
super(ResNext, self).__init__()
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, ceil_mode=True)
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.AvgPool2d(7)
self.fc = nn.Linear(512 * block.expansion*9, num_classes)
for m in self.modules():
if isinstance(m, nn.Conv2d):
n = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
m.weight.data.normal_(0, math.sqrt(2. / n))
elif isinstance(m, nn.BatchNorm2d):
m.weight.data.fill_(1)
m.bias.data.zero_()
def _make_layer(self, block, planes, blocks, stride=1):
downsample = None
#-------------------------------------------------------------------#
# 当模型需要进行高和宽的压缩的时候，就需要用到残差边的downsample
#-------------------------------------------------------------------#
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))
self.inplanes = planes * block.expansion
for i in range(1, blocks):
layers.append(block(self.inplanes, planes))
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)
print(x.shape)
x = self.fc(x)
return x
arch_settings = {
'resnext50': (Bottleneck_50, (3, 4, 6, 3)),
'resnext101': (Bottleneck_101, (3, 4, 23, 3))
}
def resnext(depth,pretrained = False):
if depth not in arch_settings:
raise KeyError(f'invalid depth {depth} for resnet')
Bottleneck,stage_blocks = arch_settings[depth]
model = ResNext(Bottleneck, stage_blocks)
if pretrained and depth=="resnext50":
resnext50 = models.resnext50_32x4d(pretrained=True)
pretrained_dict = resnext50.state_dict()
model_dict = model.state_dict()
load_key, no_load_key, temp_dict = [], [], {}
for k, v in pretrained_dict.items():
if k in model_dict.keys() and np.shape(model_dict[k]) == np.shape(v):
temp_dict[k] = v
load_key.append(k)
else:
no_load_key.append(k)
model_dict.update(temp_dict)
model.load_state_dict(model_dict)
print("load over")
print("\nSuccessful Load Key:", str(load_key)[:500], "……\nSuccessful Load Key Num:", len(load_key))
print("\nFail To Load Key:", str(no_load_key)[:500], "……\nFail To Load Key num:", len(no_load_key))
elif pretrained and depth == "resnext101":
resnext101 = models.resnext101_32x8d(pretrained=True)
pretrained_dict = resnext101.state_dict()
model_dict = model.state_dict()
load_key, no_load_key, temp_dict = [], [], {}
for k, v in pretrained_dict.items():
if k in model_dict.keys() and np.shape(model_dict[k]) == np.shape(v):
temp_dict[k] = v
load_key.append(k)
else:
no_load_key.append(k)
model_dict.update(temp_dict)
model.load_state_dict(model_dict)
# print("load over")
print("\nSuccessful Load Key:", str(load_key)[:500], "……\nSuccessful Load Key Num:", len(load_key))
print("\nFail To Load Key:", str(no_load_key)[:500], "……\nFail To Load Key num:", len(no_load_key))
return model
if __name__=="__main__":
model = resnext('resnext50',pretrained=False).cuda()
model = model.cuda()
summary(model,(3,640,640))