-
Notifications
You must be signed in to change notification settings - Fork 267
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
Submitting SegDataGenerator to Keras? #8
Comments
Of course it'll be great if you can help submitting it. But you need to make sure it can support 'th' dim_ordering correctly, which I've never checked. And I guess it's also needed to fix lots of my dirty codes...... Anyway it will be really appreciated if you can help, at least this can give an example for those who wanna use keras to do image segmentation. |
Just to update, I've been doing some code cleanup and working to get all the pieces in so an example can also be included in keras-contrib. I've submitted pull requests for:
Eventually, the code, particularly SegDataGenerator, will likely need to be moved to most likely an The model creation code might then go into applications, and train.py would go into a single file with multiple command line options in https://github.com/farizrahman4u/keras-contrib/tree/master/examples. It may be possible to handle the tensorflow vs theano part later on, as long as tf isn't called directly in anything under keras_contrib that is not in the tensorflow backend. |
Update: François Chollet, Keras' author, said he is interested in directly incorporating dense prediction/FCN into the Keras API, so I'm seeking suggestions/feedback at keras-team/keras#6538 |
@Irtza I'd appreciate if your question could be in a separate github issue. A setup.py pull request would be welcome, or just add it to your PYTHONPATH. |
Okay I will add more details on the roadblocks while using the segDataGenerator in another issue on this repo |
Would you be interested in submitting or giving permission to submit SegDataGenerator to keras?
It would first go through keras-contrib as per the Keras instructions for pull requests.
I wouldn't mind doing the integration, I can make sure credit is properly given as well.
The text was updated successfully, but these errors were encountered: