TTA@0.8

[endeavorXx]

I do not understand the meaning of TTA@0.8 in the context of whole dataset?

Let's consider a simple case, I have hard preds labels "0" and "1".
What do you mean by TTA@0.8 in the context of the whole dataset which contains multiple videos?
How we can calculate TTA@0.8 as every video will have it's own recall and precision.