web3-202226701039/DensenetModels.py

import os
import numpy as np

import torch
import torch.nn as nn
import torch.backends.cudnn as cudnn
from CBAM import CBAM
import torchvision.transforms as transforms
from torch.utils.data import DataLoader
from sklearn.metrics import roc_auc_score

import torchvision

# class FocalLoss(nn.Module):
#     def __init__(self, alpha=1, gamma=2, logits=True, reduce=True):
#         super(FocalLoss, self).__init__()
#         self.alpha = alpha
#         self.gamma = gamma
#         self.logits = logits
#         self.reduce = reduce

#     def forward(self, inputs, targets):
#         if self.logits:
#             BCE_loss = torch.nn.functional.binary_cross_entropy_with_logits(inputs, targets, reduction='none')
#         else:
#             BCE_loss = torch.nn.functional.binary_cross_entropy(inputs, targets, reduction='none')
#         pt = torch.exp(-BCE_loss)
#         F_loss = self.alpha * (1 - pt) ** self.gamma * BCE_loss
#         if self.reduce:
#             return torch.mean(F_loss)
#         else:
#             return F_loss
class FocalLoss(nn.Module):
    def __init__(self, alpha=None, gamma=2, logits=True, reduce=True):
        super(FocalLoss, self).__init__()
        if alpha is None:
            self.alpha = 1.0
        else:
            self.alpha = alpha
        self.gamma = gamma
        self.logits = logits
        self.reduce = reduce

    def forward(self, inputs, targets):
        if self.logits:
            BCE_loss = torch.nn.functional.binary_cross_entropy_with_logits(inputs, targets, reduction='none')
        else:
            BCE_loss = torch.nn.functional.binary_cross_entropy(inputs, targets, reduction='none')

        pt = torch.exp(-BCE_loss)

        # 应用类别权重 alpha
        if isinstance(self.alpha, torch.Tensor):
            self.alpha = self.alpha.to(inputs.device)
            alpha_factor = self.alpha.unsqueeze(0)  # 调整形状以匹配输入 (batch_size, num_classes)
        else:
            alpha_factor = self.alpha

        F_loss = alpha_factor * (1 - pt) ** self.gamma * BCE_loss

        if self.reduce:
            return torch.mean(F_loss)
        else:
            return F_loss
class DenseNet121(nn.Module):

    def __init__(self, classCount, isTrained):
        super(DenseNet121, self).__init__()

        self.densenet121 = torchvision.models.densenet121(pretrained=isTrained)

        kernelCount = self.densenet121.classifier.in_features

        self.densenet121.classifier = nn.Sequential(
            nn.Linear(kernelCount, classCount))

    def forward(self, x):
        x = self.densenet121(x)
        return x
class SELayer(nn.Module):
    def __init__(self, channel, reduction=16):
        super(SELayer, self).__init__()
        self.avg_pool = nn.AdaptiveAvgPool2d(1)
        self.fc = nn.Sequential(
            nn.Linear(channel, channel // reduction, bias=False),
            nn.ReLU(inplace=True),
            nn.Linear(channel // reduction, channel, bias=False),
            nn.Sigmoid()
        )

    def forward(self, x):
        b, c, _, _ = x.size()
        y = self.avg_pool(x).view(b, c)
        y = self.fc(y).view(b, c, 1, 1)
        return x * y.expand_as(x)

# 修改后的DenseNet121，增加SE模块
# class DenseNet121(nn.Module):
#     def __init__(self, num_classes, pretrained=True):
#         super(DenseNet121, self).__init__()
#         # 加载预训练的DenseNet121模型
#         self.densenet = torchvision.models.densenet121(weights="IMAGENET1K_V1" if pretrained else None)
#         self.se = SELayer(self.densenet.features[-1].num_features)
#         kernel_count = self.densenet.classifier.in_features
#         self.densenet.classifier = nn.Linear(kernel_count, num_classes)
#
#     def forward(self, x):
#         # 特征提取
#         x = self.densenet.features(x)
#         # 使用SE模块
#         x = self.se(x)
#         # 全局平均池化以确保形状匹配
#         x = torch.nn.functional.adaptive_avg_pool2d(x, (1, 1))
#         # 展平特征
#         x = torch.flatten(x, 1)
#         # 分类层
#         x = self.densenet.classifier(x)
#         return x

# class DenseNet121(nn.Module):
#     def __init__(self, num_classes, pretrained=True):
#         super(DenseNet121, self).__init__()
#         self.densenet121 = torchvision.models.densenet121(
#             weights="IMAGENET1K_V1" if pretrained else None)
#         kernel_count = self.densenet121.classifier.in_features

#         self.cbam = CBAM(kernel_count)

#         # 移除 Sigmoid 激活
#         self.densenet121.classifier = nn.Linear(kernel_count, num_classes)

#     def forward(self, x):
#         x = self.densenet121.features(x)
#         x = self.cbam(x)
#         x = torch.nn.functional.adaptive_avg_pool2d(x, (1, 1))
#         x = torch.flatten(x, 1)
#         x = self.densenet121.classifier(x)
#         return x

class DenseNet169(nn.Module):

    def __init__(self, classCount, isTrained):
        super(DenseNet169, self).__init__()

        self.densenet169 = torchvision.models.densenet169(pretrained=isTrained)

        kernelCount = self.densenet169.classifier.in_features

        self.densenet169.classifier = nn.Sequential(
            nn.Linear(kernelCount, classCount), nn.Sigmoid())

    def forward(self, x):
        x = self.densenet169(x)
        return x


class DenseNet201(nn.Module):

    def __init__(self, classCount, isTrained):
        super(DenseNet201, self).__init__()

        self.densenet201 = torchvision.models.densenet201(pretrained=isTrained)

        kernelCount = self.densenet201.classifier.in_features

        self.densenet201.classifier = nn.Sequential(
            nn.Linear(kernelCount, classCount), nn.Sigmoid())

    def forward(self, x):
        x = self.densenet201(x)
        return x

class ResNet50(nn.Module):
    def __init__(self, classCount, isTrained=True):
        super(ResNet50, self).__init__()
        self.resnet50 = torchvision.models.resnet50(weights="IMAGENET1K_V1" if isTrained else None)
        self.se = SELayer(self.resnet50.layer4[-1].conv3.out_channels)
        kernelCount = self.resnet50.fc.in_features
        self.resnet50.fc = nn.Linear(kernelCount, classCount)

    def forward(self, x):
        # 特征提取
        x = self.resnet50.conv1(x)
        x = self.resnet50.bn1(x)
        x = self.resnet50.relu(x)
        x = self.resnet50.maxpool(x)

        x = self.resnet50.layer1(x)
        x = self.resnet50.layer2(x)
        x = self.resnet50.layer3(x)
        x = self.resnet50.layer4(x)
        # 使用SE模块
        x = self.se(x)
        # 全局平均池化以确保形状匹配
        x = torch.nn.functional.adaptive_avg_pool2d(x, (1, 1))
        # 展平特征
        x = torch.flatten(x, 1)
        # 分类层
        x = self.resnet50.fc(x)
        return x