Support Pytorch & Tensorflow Formats on Datasets

[PJEstrada]

This branch will add 3 main features:

• DatasetIterator class: will allow all dataset objects to be iterable so we can directly get all the instances on each file in a dataset with a simple forloop.
• pytorch dataset class: builds on to of the base iterator and create a pytorch dataset object.
• tensorflow dataset class: creates a tf dataset object.

This version will limit to only bbox and polygon instance types. We will also add the ability to slice a dataset with the query engine an iterate it directly.

[PJEstrada]

This branch is ready for review: some things I want to discuss related to the to_tensorflow():

I've found lots of ways that people ingest data to model so still not 100% sure on what would be a good standard to follow. For now I'm returning a tf.data.example object with the file's instances and image data. Please let me know if you think this is a right direction.

Other from that, I think this branch should be ready:

• Added Sliced dataset class. For queries in dataset
• Added the to_pytorch() function
• Added the to_tensorflow() function
• Addded the slice() function
• Added the explore() function

[anthony-chaudhary]

This is a good direction! Let's touch on some specifics on call :)

…

On Wed, Aug 25, 2021 at 8:31 AM PJEstrada ***@***.***> wrote: This branch is ready for review: some things I want to discuss related to the to_tensorflow(): I've found lots of ways that people ingest data to model so still not 100% sure on what would be a good standard to follow. For now I'm returning a tf.data.example object with the file's instances and image data. Please let me know if you think this is a right direction. Other from that, I think this branch should be ready: - Added Sliced dataset class. For queries in dataset - Added the to_pytorch() function - Added the to_tensorflow() function - Addded the slice() function - Added the explore() function — You are receiving this because your review was requested. Reply to this email directly, view it on GitHub <#4 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AEJ6DJFH3TGVNKPNYESXHS3T6UEEFANCNFSM5B2XXJZA> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&utm_campaign=notification-email> .