ResNet_for_CC.py

import torch
import torch.nn as nn
import torchvision.models as models


class ResClassifier(nn.Module):
    def __init__(self, class_num=14):
        super(ResClassifier, self).__init__()
        self.fc1 = nn.Sequential(
            nn.Linear(128, 64),
            nn.BatchNorm1d(64, affine=True),
            nn.ReLU(inplace=True),
            nn.Dropout()
        )
        self.fc2 = nn.Sequential(
            nn.Linear(64, 64),
            nn.BatchNorm1d(64, affine=True),
            nn.ReLU(inplace=True),
            nn.Dropout()
        )
        self.fc3 = nn.Linear(64, class_num)

    def forward(self, x):
        fc1_emb = self.fc1(x)
        fc2_emb = self.fc2(fc1_emb)
        logit = self.fc3(fc2_emb)
        return logit


class CC_model(nn.Module):
    def __init__(self, num_classes=14):
        super(CC_model, self).__init__()
        self.num_classes = num_classes
        self.model_resnet = models.resnet50(weights='ResNet50_Weights.DEFAULT')

        # Modify final layers
        num_ftrs = self.model_resnet.fc.in_features
        self.model_resnet.fc = nn.Identity()  # Remove ResNet's default final layer
        self.classification_fc = nn.Linear(num_ftrs, num_classes)  
        self.dr = nn.Linear(num_ftrs, 128)  # Feature reduction (for embeddings)
        self.fc1 = ResClassifier(num_classes)
        self.fc2 = ResClassifier(num_classes)

    def forward(self, x):
        feature = self.model_resnet(x)
        class_logits = self.classification_fc(feature)  # Correct classification output
        return class_logits  # Ensure output shape is [batch_size, 14]