What is the difference between clip accuracy and video accuracy？

[liubc17]

I just use the test.list to test the trained model. It has 3351 lines and I give every line a bool (true or false). Finally, I add them to calculate the final result. Is it a clip accuracy or video accuracy? Hope for your help. Thanks. :-)