Manufacturing Idea – Could Machine Learning Super Resolution Help Manufacturing?

My idea is simple.

The idea is that you learn a model to get better resolution by probable imagination. Or at least there are some examples on github.

Then could this be tested in a manufacturing situation? To see if performance or accuracy could be improved.

To improve batteries for maybe?