How to read ImageNet dataset, val or test file?

vision
1

I am getting a RuntimeError. Recently I downloaded the ImageNet dataset from Kaggle and it is a large zip file. I just want to evaluate a trained dataset, but I can’t read the eval file.

path = "/home/me/Desktop/IMGNetDataset/imagenet-object-localization-challenge/ILSVRC/Data/CLS-LOC/val"


batch_size = 128

workers = 0

transform = transforms.Compose([ transforms.Resize(256),
                                    transforms.CenterCrop(224),
                                    transforms.ToTensor(),
                                    transforms.Normalize(mean=[0.485, 0.456, 0.406], std=[0.229, 0.224, 0.225])
                                  ])

preprocessing = dset.ImageNet(root=import_dataset,
                                                    transform = transform )
                                
test_sampler = SequentialSampler(preprocessing)


dataloader = DataLoader(preprocessing, batch_size=batch_size, sampler= test_sampler,
                               shuffle=False, num_workers=workers)

There is not such file ILSVRC2012_devkit_t12.tar.gz in the downloaded dataset.

RuntimeError: The archive ILSVRC2012_devkit_t12.tar.gz is not present in the root directory or is corrupted. You need to download it externally and place it in /home/me/Desktop/IMGNetDataset/imagenet-object-localization-challenge/ILSVRC/Data/CLS-LOC/.
2

I had a similar issue before. Anyways, The error message indicates that the file ILSVRC2012_devkit_t12.tar.gz is missing or corrupted. Try Downloading the file from the original source and place it in the proper directory.

3

Thanks, but now works only with the “train” file and doesn’t recognize any “val” file.


preprocessing = dset.ImageNet(root=datapath,
                                  split="val",
                                  transform= transform)


FileNotFoundError: Couldn't find any class folder in /home/me/Desktop/IMGNetDataset/imagenet-object-localization-challenge/ILSVRC/Data/CLS-LOC/val.

The file “val” contains ILSVRC2012_val_00000001.JPEG , I don’t know if is it in the right form. The train file contains subfolders

4

I checked the ImageNet dataset in google Colab. There is no problem. That file name is fine.
I used the links below to download “val”:

!wget https://image-net.org/data/ILSVRC/2012/ILSVRC2012_img_train.tar --no-check-certificate
and
!wget https://image-net.org/data/ILSVRC/2012/ILSVRC2012_img_val.tar --no-check-certificate

Screenshot (434)
Screenshot (434)1541×834 239 KB

1 Like
5

Yes it works now, for train and val, what about test file? There is no option in PyTorch for splitting also the test file.