[data request] tiny-imagenet

• Name of dataset: tiny-imagenet
• URL of dataset: http://cs231n.stanford.edu/tiny-imagenet-200.zip
• License of dataset: Not mentioned
• Short description of dataset and use case(s): Tiny ImageNet Challenge is a similar challenge as ImageNet with a smaller dataset but with less image classes. It contains 200 image classes, a training dataset of 100, 000 images, a validation dataset of 10, 000 images, and a test dataset of 10, 000 images. All images are of size 64×64. It was created for CS231N course at Stanford.

Folks who would also like to see this dataset in tensorflow/datasets, please thumbs-up so the developers can know which requests to prioritize.

And if you'd like to contribute the dataset (thank you!), see our guide to adding a dataset.

I have already implemented the support for this dataset. Please see here -
https://github.com/ksachdeva/tiny-imagenet-tfds

I developed it as a standalone repository to test out the tensorflow_datasets framework for private datasets (and it does work .. thanks for the great work here). I have also published an article here that explains the motivations behind tensorflow_datasets and details on how one can use it for their own private dataset.

Link to the article on medium:
https://towardsdatascience.com/a-unified-method-for-downloading-extracting-processing-using-datasets-7482a3b27aff

If you are okay adding the support for this dataset in the collection, I will send the pull requests (and mostly update the code) to be compliant with the contribution guides.

Start with docs/add_dataset.md and compare the linked standalone tiny-imagenet-tfds implementation with the contribution requirements. Confirm the dataset details and license status, then prepare compliant tensorflow_datasets support; done means the dataset is added according to the guide and its download, splits, and metadata are represented.