Which dataloader is good for labels saved in csv file?

[John1231983]

Hello all, I found many kinds of data loader (i.e ImageFolder, …) in pytorch. For my problem, I have a training data that organizer as:

animal

|-111.jpg
|-121.jpg
|-131.jpg
...

and a csv file store the label information

111.jpg, 1
121.jpg,0
131.jpg,1

So, what kind of dataloader is good for me? Could you show some example code

[Nikronic (Nikan Doosti)]

I have asked something similar to your question and I find out writing a custom dataset based on Dataset abstract class is the best idea and fastest!

This is a sample as an answer to my question:

Custom Dataset with some preprocessing

When should I do this based on Dataset class? Should I do my processing step in __getitem__() ? If yes, would it be parallel and fast? Yes. the __getitem__ calls are the ones run in parallel. Here’s some example: class CelebaDataset(Dataset): """Custom Dataset for loading CelebA face images""" def __init__(self, txt_path, img_dir, transform=None): df = pd.read_csv(txt_path, sep=" ", index_col=0) self.img_dir = img_dir self.txt_path = txt_path …

And if you are in full implemented of my work, here is the link to my Github of a simple CNN for edge detection on some custom dataset with custom loss and custom transforms.

And my dataset structure is exactly same as you. So I hope it helps you.
Thanks for any star