In my experience with computer science conferences, I have encountered two general classes of award processes:
- Reviewers nominate for an award, and then the organizers consider and choose between the nominees.
- After reviews have been submitted, the organizers consider the top N papers by rating and choose between them.
Sometimes the decision is made entirely from the paper and reviews, while in other cases presentation at the conference is taken into account as well. In all of these cases, however, notice that it is the organizers and not the reviewers who actually decide the award, and the authors are generally not blinded to the organizers. If the first procedure is followed and the conference happens to be double-blind, then nominations will be made blind, but the decision is still typically made with knowledge of the authors.