Apply Torchvision Transforms to a numpy array

Question

I wrote the following code:

transform = transforms.Compose([
    transforms.ToPILImage(),
    transforms.RandomResizedCrop(224),
    transforms.RandomHorizontalFlip(),
    transforms.ToTensor(),
    transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225]),
])


train_dataset = torch.utils.data.TensorDataset(torch.from_numpy(X_train).float(), torch.from_numpy(y_train).float())
val_dataset = torch.utils.data.TensorDataset(torch.from_numpy(X_val).float(), torch.from_numpy(y_val).float())


# Define the dataloaders
train_loader = torch.utils.data.DataLoader(train_dataset, batch_size=32, shuffle=True, num_workers=4)
val_loader = torch.utils.data.DataLoader(val_dataset, batch_size=32, shuffle=False, num_workers=4)

I would like to apply the transform compose to my dataset (X_train and X_val) which are both numpy array. How can I apply transform to augment my dataset and normalize it. Should I apply it before the model training or during model training?

Answer 1

You can simply add the transformers to your training pipeline:

train_transformed = transform(torch.from_numpy(X_train).float())
val_transformed = transform(torch.from_numpy(X_val).float())

train_dataset = torch.utils.data.TensorDataset(train_transformed, torch.from_numpy(y_train).float())
val_dataset = torch.utils.data.TensorDataset(val_transformed, torch.from_numpy(y_val).float())

This way the transform is applied during data loading which is before training.