How can I load the (labels of the) imagenet val dataset?

I would like to work with the validation split from huggingface dataset. How can I do that?

I downloaded the validation images here val_images.tar.gz and stored them in my a local folder. I also downloaded the files, dataset_infos.json and

When I then load the dataset with

ds = load_dataset('/home/imagenet/val/', split='validation')

How can I iterate of this dataset? It seems to me that it only contains as features [‘image’] but not the labels.