[Info Need] Clean implementation of Object Detection models in keras

I need some information about the clean and best practice implementation of object detection models with tf.keras. Here is one of the two implementations for example:

Any more recommendations? Please note, it should be a high-quality implementation and trusted. FYI, I can also use TF-Model but no thanks, maybe later. :roll_eyes:

You can find some refactored model in the model garden at:

1 Like

@Bhack Thanks. I went through these beta directories before. I think it’s not properly documented, there’s a lack of information in the read-me file (how to do…). Again it’s a beta version. Is there any separate resource where people use this beta version?

And curiously asking, I’ve read somewhere about keras-cv; what would be the relation or boundaries between these two repositories?

Yes the beta It is still WIP so documentation Will come soon or later.

For keras-cv you can see:

I’m not 100% sure, but I believe the plan was: put the reusable components (layers, losses, metrics) in keras-cv, and put the model implementations in tensorflow/models. But It’s not clear what the status of this plan is at the moment.