Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

'Created a Keras Based Iterator, correct a previous bug from master' #105

Open
wants to merge 3 commits into
base: master
Choose a base branch
from

Conversation

tchaton
Copy link

@tchaton tchaton commented Apr 13, 2018

I have observed that keras_generator was randomly sampling in range(0, len (self.augmentor_images) -1) which doesn't actually a full iteration over the data. I modified the code to add Keras Iterator support.
I also added a with_indexes with can be useful, if you want to track the images coming out of the generator.

@mdbloice
Copy link
Owner

Hi @tchaton, thanks a lot, I will take a close look at that as soon as I get the chance, most likely on Monday.
Thanks again!

@tchaton
Copy link
Author

tchaton commented Apr 13, 2018

I have forgot to import threading inside Augmentor/Pipeline.py
I will also add some support for ImgAug library. If we are able to rely on it over the operations (very simple to add, already have my custom ones) and handle pipeline using this library, it will simply everthing.
My main issue is operations are made over Pil Images, where it should be numpy array.

@mdbloice
Copy link
Owner

I am not sure if relying on ImgAug makes that much sense... well, do you mean that Augmentor should provide an API so that it could interact with ImgAug easily, or do you mean that it would use ImgAug as a dependency?

@tchaton
Copy link
Author

tchaton commented Apr 13, 2018 via email

@tchaton
Copy link
Author

tchaton commented Apr 14, 2018

I have corrected it. It should work fine.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants