At the moment the system does not support auto labelling, but we're trying to work around the issue by importing dataset which already contains predicted annotations. Unfortunately all entries in this dataset are marked as completed. That means it's impossible for annotator to start the work and see his progress.

I think in use cases like this it would make sense to mark the item as completed only if it contains annotations AND was reviewed by user (opened and user clicked next). In this use case it might be useful to have a button to "skip" entry in dataset. Not to mark item as completed accidentally.