In order to validate the output from inference on custom trained models, we need to have a pipeline that allows for simple human validation. We should have a labelstudio (and potentially another validation software) compatible upload script to easily validate the output of ML inference.