model.py

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

class CNNLSTMClassifier(nn.Module):
    def __init__(self, hidden_dim=128, num_classes=2):
        super().__init__()
        resnet = models.resnet18(weights=models.ResNet18_Weights.DEFAULT)
        self.cnn = nn.Sequential(*list(resnet.children())[:-1])
        self.cnn_out_dim = 512
        self.lstm = nn.LSTM(self.cnn_out_dim, hidden_dim, batch_first=True)
        self.fc = nn.Linear(hidden_dim, num_classes)

    def forward(self, x):
        B, T, C, H, W = x.shape
        x = x.view(B * T, C, H, W)
        with torch.no_grad():
            cnn_out = self.cnn(x).view(B, T, -1)
        lstm_out, _ = self.lstm(cnn_out)
        return self.fc(lstm_out[:, -1, :])