The task is very simple. Android app will load a pytorch model and a data file and show the output in a textbox or any suitable format.
I tried to follow the projects there but each time my app crashes.
I think this is a very simple task. Just upload models and data and show output in android app.
I looked at many github repositories but most of them are 2 years ago at least.
Is there any recent Open Source Code available that anyone can share.
In recent android studio version, you can use your pytorch (.pt) model into your application. But to do that you have to convert your pytorch model to onnx which enables us to convert framework models to other framework model and then convert your onnx model to TensorFlow Lite model.
1- convert pytorch model to onnx
2- convert onnx model to tflite model
3- And add tflite model into your app
android docs is here : Build a handwritten digit classifier app with TensorFlow Lite | Android Developers
Do you know any resource where they have implemented the full pipeline?
It can be a very simple app. But from pytorch model to android app.
Thanks.
What do you mean by saying “implemented full pipeline” ? If I am not wrong, you want to use your pytorch model in your android app right?
To do that you have to convert your pytorch model to onnx and then convert from onnx to Tflite. You can see many resources about that conversation If you google it.