KeyError when using torch.utils.data.Subset for specific dataset but not cifar

vision
#1

I’m getting the following error:

KeyError: tensor(0)

This happens after running the following for loop:

train_dataset = CovidDataset(train, transform=torchvision.transforms.ToTensor())
print(f"Original Train dataset length: {train_dataset.__len__()}")
indices = torch.arange(3000)
train_dataset = torch.utils.data.Subset(train_dataset, indices)
print(f"New Train dataset length: {train_dataset.__len__()}")

for idx, (data, img) in enumerate(train_dataset):
    print(idx)

However when I try it on CIFAR dataset, it works fine:

# import the required modules
import torch
import torchvision
from torchvision.datasets import CIFAR10
from collections import Counter

trainset = CIFAR10(root='./data', train=True, download=True, transform=torchvision.transforms.ToTensor())

indices = torch.arange(3000)
trainset = torch.utils.data.Subset(trainset, indices)

for idx, (data, img) in enumerate(trainset):
    print(idx)

Why does it work on CIFAR dataset but not on my dataset?

#2

Full Code:

Link to dataset: COVIDx CT | Kaggle

import torch
import torchvision
import matplotlib.pyplot as plt
import pandas as pd
import numpy as np
import PIL
import os

# Convert .txt to .csv
TXT_FILE_PATH = r"C:\Users\ASUS\Machine Learning\data"
train = pd.read_csv(TXT_FILE_PATH + r"\train_COVIDx-CT.txt", sep=" ")
train.columns=['filename', 'label', 'xmin','ymin','xmax','ymax']
train=train.drop(['xmin', 'ymin','xmax', 'ymax'], axis=1 )

# Put file path to start of image so that it can be loaded
IMG_FILE_PATH = "C:\\Users\\ASUS\\Machine Learning\\data\\images\\"
train['filename'] = IMG_FILE_PATH + train['filename']

# Hyperparameters
BATCH_SIZE = 32
IMG_SIZE = (224,224)
learning_rate = 0.001
num_epochs = 10
in_channels = 1

# Custom Dataset
class CovidDataset(torch.utils.data.Dataset):
    def __init__(self, file_type, transform=None):
        self.file_type = file_type
        self.transform = transform
    def __getitem__(self, idx):
        img_name = self.file_type['filename'][idx]
        img = PIL.Image.open(img_name)
        label = self.file_type['label'][idx]
        if self.transform:
            img = self.transform(img)
        return img, label
    
    def __len__(self):
        return len(self.file_type)
    
train_dataset = CovidDataset(train, transform=torchvision.transforms.ToTensor())
print(f"Original Train dataset length: {train_dataset.__len__()}")
indices = torch.arange(3000)
train_dataset = torch.utils.data.Subset(train_dataset, indices)
print(f"New Train dataset length: {train_dataset.__len__()}")

for idx, (data, img) in enumerate(train_dataset):
    print(idx)