Documentation/Tutorial for custom Active Learning using a tf 2 vision model

[kadijahassanali]

Is your feature request related to a problem? Please describe.
I have tried multiple strategies to use a custom object detection model in Active Learning. These include trying to convert both keras and tf saved_models to tfjs format using the tfjs converter. The conversions have been successful but none of them are successfully loaded to VOTT.
Describe the solution you'd like
A solution/tutorial on how to implement a custom Active Learning model with a tf release < 2 years old.
Describe alternatives you've considered
Switching to another labelling tool