Dataset separation and models

[wghrayeb]

Dear Joseph and team,

This project is great reference to learn and think about the topic of lip reading. Thank you!

I was wondering what do you mean by the separation of datasets to "nano", "small", "medium", etc? is it automatically produced by any script or manually categorized?

Are there in the Repo any model that was trained on any of those dataset that one can use to see how the net works?

Thanks in advance.

[joseph-zhong]

The terms "nano", "small", etc were our own semantics to mean the size of the dataset and really was for convenience in quickly prototyping ideas on smaller datasets before moving onto larger ones.

Unfortunately we haven't had the opportunity of releasing trained models

[wghrayeb]

I see. thank you for explanation.

The pipeline looks very interesting, I'm wondering if you ended up writing a paper? or publishing results that shows the promise of your work?