What would be minimum SDK(Software Development Kit) level in Android Studio for making a Machine Learning model based application

Question

Basically, I have to make ML-based application in Android Studio. I already make and trained my ML model and converted it into the TensorFlow Lite model. But when it comes to the android studio while configuring my project there's an option for minimum SDK. As far as I know, you have to select the minimum SDK level so that your application can run on every android version. In that case, how can I ensure that ML-based application will work fine on the older android version? This is the reason I am asking what would be the minimum SDK level for ML-based application.

Answer 1

minimum SDK level is used to determine minimum API you want your application to support. It doesn't quite related to your tflite model.

I think what's most important is that you pull in correct tflite runtime which has all ops that your model uses.

Not sure whether you model has metadata. If it does, maybe you can use ml model binding so you don't need write too many boilerplate code.