Is Overtraining a Good Assessment?

It's hard to correct for input data that is that has some kind of bias or some kind of issue with it. And then secondly is, is avoiding things like over training right? We're doing at the moment with unity frameworks training our computer vision within a VR environment and it just has no reference to what it's looking at because it's kind of saying, here's a picture of me walking do more something else. This is what it actually looks like is correct most of the time.