demo site for keras-cv layers

[pure-rgb]

there are lots of preprocessing and augmentation layers that are hard to understand. can keras offer some like https://demo.albumentations.ai/ - an interactive site.

[jbischof]

Agree this would be cool, but we don't have the bandwidth unfortunately. Is this something you could contribute @pure-rgb?

[pure-rgb]

Yes, I would love to contribute. This should be live in keras-io.
cc. @fchollet

[jbischof]

@pure-rgb does this proposal require any changes to the codebase? What if you make a demo site for a handful of augmentations so I can get feedback from the team? Thank you!

[IMvision12]

hey @pure-rgb are u working on this?

[pure-rgb]

@IMvision12 I'm not free at the moment, but hopefully I will catch up ASAP. If you want you can start working on it, and later we can collaborate for sure.

[IMvision12]

Hey @pure-rgb thanks I have started, and almost completed with frontend and backend just solving some errors

[james77777778]

Hey @IMvision12 @pure-rgb @jbischof
I would like to share some code that could be helpful in creating a demo site using Streamlit.
https://github.com/james77777778/keras-aug/tree/main/streamlit_app

• app.py contains the main logic and layout of the site
• layers_config.py contains the configuration of the layers

Live demo on Streamlit cloud:
https://james77777778-keras-aug-streamlit-appapp-mxd7v1.streamlit.app/

As KerasAug is built on KerasCV, it should be easy to follow the implementation.

P.S. I'm willing to improve the preprocessing layers in KerasCV if the maintainers consider the implementation in KerasAug to be beneficial

[IMvision12]

@james77777778 Thank you for sharing this; I was planning to develop a website using HTML, CSS, JS, ReactJs, and Flask, but I believe using Streamlit will save time and be simpler.

[IMvision12]

@james77777778 BTW, do you intend to add bbox augmentation support to your website?

[james77777778]

@james77777778 BTW, do you intend to add bbox augmentation support to your website?

Yes, it's part of my plan, but it might take some time.

[IMvision12]

https://imvision12-kerascv-demo-app-15jw4n.streamlit.app/

did some changes, but major code was taken from https://github.com/james77777778/keras-aug/tree/main/streamlit_app
Thanks to @james77777778

Working on adding support for bbox augmentation!

@ianstenbit @jbischof

[jbischof]

This is super cool! I think the option to add a relevant box/mask to the image to see that transformed as well would definitely add to the utility.

[ianstenbit]

+1 -- this is awesome! Just played around with it a little bit. If this is easy to host we could maybe add it to KerasIO or something like that and link to it from our docs.

Great stuff 😄

[IMvision12]

Added bounding box support! currently, it supports "xywh" and "xyxy" formats!

[IMvision12]

Added segmentation support :) !

[IMvision12]

@jbischof @ianstenbit Let me Know where should i add this app, in KerasCV by creating a new folder named streamlit_app or Keras.io

[ianstenbit]

@jbischof @ianstenbit Let me Know where should i add this app, in KerasCV by creating a new folder named streamlit_app or Keras.io

How much effort does it take to deploy this locally / host it?
If it's something we could serve statically, would GitHub pages work as an option? (I don't really know anything about streamlit or the details of how serving this works).

Otherwise we can also just include a link to this in our README / docs.

The mask support looks awesome 😄

[innat]

I think this should be added to the Keras-io website. This way it will reach and attract more users.

[jbischof]

@innat we don't know how to do this and are asking

[IMvision12]

@ianstenbit @innat @jbischof We cannot host the app on GitHub pages reason: https://discuss.streamlit.io/t/hosting-streamlit-on-github-pages/356/3.

I'm not familiar with the hosting details of keras.io, so I'm uncertain if it is possible to deploy this application on keras.io.

Suggestion: we can create a folder in kerascv itself and host, that's how I have hosted the app from my repo: https://github.com/IMvision12/kerascv-demo

Details of hosting streamlit app : https://blog.streamlit.io/host-your-streamlit-app-for-free/

[pure-rgb]

I'm not familiar with the hosting details of keras.io, so I'm uncertain if it is possible to deploy this application on keras.io.

@fchollet

[sachinprasadhs]

Thanks for reporting the issue! We have consolidated the development of KerasCV into the new KerasHub package, which supports image, text, and multi-modal models. Please read keras-team/keras-hub#1831. KerasHub will support all the core functionality of KerasCV.

KerasHub can be installed with !pip install -U keras-hub. Documentation and guides are available at keras.io/keras_hub.

With our focus shifted to KerasHub, we are not planning any further development or releases in KerasCV. If you encounter a KerasCV feature that is missing from KerasHub, or would like to propose an addition to the library, please file an issue with KerasHub.

[innat]

@sachinprasadhs

If you encounter a KerasCV feature that is missing from KerasHub, or would like to propose an addition to the library, please file an issue with KerasHub.

Firstly, identifying which feature is required or missing can be done effectively, either by practitioners or the Keras team. Tickets in keras-cv were created only when such features were NOT available or when a specific need was encountered. Commenting after a significant delay and requesting the recreation of the same issue is neither efficient nor appropriate. Kindly move this ticket (and any other similar tickets with the same context) from keras-cv to keras-hub.