Skip to content

Getting Started Tutorial #68

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

Open
alexdriedger opened this issue Jun 7, 2020 · 7 comments
Open

Getting Started Tutorial #68

alexdriedger opened this issue Jun 7, 2020 · 7 comments
Assignees

Comments

@alexdriedger
Copy link

Describe the documentation issue

Before the java keras development was moved to this repo, it had a great getting started guide. That would be great to be easily discoverable in this repo.

(A getting started guide for importing models from python would be great too. I have a feeling that's a pretty common use case for using tensorflow in java)

@karllessard
Copy link
Collaborator

karllessard commented Jun 17, 2020

Hi @alexdriedger , sorry for the delay, last days were a bit crazy for me,

I totally agree that we should start focusing on a simple tutorial explaining different steps for doing inference in TF Java from a saved model (probably trained in Python), this is the most common use case of our library right now. Would you be interested doing it? We also had a few discussions with Google to see if we can integrate a Jupyter book with a Java kernel to write such tutorials, it might something we want to take a look at.

About Keras, it is still under heavy development and nothing has been merged so far in this repository, about this @dhruvrajan can probably provide you a better status.

@aday00
Copy link

aday00 commented Aug 17, 2020

I may be able to help here, now that model save/load seems to work (for me) from *.pb files in mainline Tensorflow-Java #100 (comment). In the near future, save/load will be different #101.

However, when I used Python Keras, I saved numpy and JSON files, not *.pb files. I did that because it seemed to most reliably work for me. Should I try to save a model as a *.pb file in Python Keras, then load that in Tensorflow-Java, and document this process? If so, kindly advise where I should put documentation? I call Tensorflow-Java from Scala, which may be convenient, but probably not as convenient as a Keras-like abstraction.

@steven-mi
Copy link

steven-mi commented Oct 2, 2020

Is there a snippet for loading a SavedModel and running inference? Does the documentation already exist?

@steven-mi
Copy link

I was able to run my SavedModel. Is there a wrapper for loading images? At the moment I am creating the input Tensorflow for my CNN manually.

@karllessard
Copy link
Collaborator

Hi @steven-mi , I invite you to take a look at this unit test to have a better idea on how to load and run saved models in Java.

This ConcreteFunction API is new and we did not updated yet our examples in this repo to reflect the latest changes. Please feel free to create a PR it you think that you may have something worth making an example of.

For image loading, there is no framework provided by our library for this at this time.

@igor-susic
Copy link

@karllessard Hi, well I would like to contribute and I guess the best way would be to write some docs :) So if it is ok you could assign me here and I will take my time to do it right.

@karllessard
Copy link
Collaborator

Thanks @igor-susic , this is greatly appreciated, documentation is definitely lacking right now!

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

No branches or pull requests

5 participants