a noise signal.mat as the resnet18' input and a clean signal as its output in python

main.py
import numpy as np
from scipy.io import loadmat
import torch
from torch.utils.data import Dataset, DataLoader
from torchvision import transforms

class SignalDataset(Dataset):
    def __init__(self, clean_file, noise_file, transform=None):
        self.clean = loadmat(clean_file)['clean_signal']
        self.noise = loadmat(noise_file)['noise_signal']
        self.transform = transform

    def __len__(self):
        return len(self.clean)

    def __getitem__(self, idx):
        clean_signal = self.clean[idx]
        noise_signal = self.noise[idx]
        sample = {'noisy': noise_signal, 'clean': clean_signal}

        if self.transform:
            sample = self.transform(sample)

        return sample

class ToTensor(object):
    def __call__(self, sample):
        noisy, clean = sample['noisy'], sample['clean']
        return {'noisy': torch.from_numpy(noisy).float(),
                'clean': torch.from_numpy(clean).float()}

# Assuming the signals are stored in MATLAB format (.mat)
clean_file = '/path/to/clean_signal.mat'
noise_file = '/path/to/noise_signal.mat'
batch_size = 32

# Define transformations to data
transform = transforms.Compose([ToTensor()])

# Load data into DataLoader
signal_dataset = SignalDataset(clean_file, noise_file, transform=transform)
dataloader = DataLoader(signal_dataset, batch_size=batch_size, shuffle=True)
1361 chars
43 lines