How to evaluate face verification performance?

I have a FACE RECOGNITION software in my PC, and I am going to evaluate its performance in VERIFICATION process.

The software permits the user to choose a similarity threshold in %, and the number of matched faces to be shown at the end of the process.

I am going to use ROC curve method to detect the value of EER (equal error rate) for the software.

My questions are:

1- How to calculate the values of "True Acceptance Rate" and "False Acceptance Rate" at each threshold?

I know that TAR=TA/(TA+FR) and FAR=FA/(FA+TR), but how should I compute each of these quantites" TA, FA ,TR, FR?

2- Should I enroll face images of N persons, or just the face images of the person under investigation?

Thanks in advance.