Are your product increments really releasable?
One of the greatest expectations of implementing Scrum is the ability to create potentially releasable increments in every cycle. This implies testing features in the same sprint (cycle) as they are being developed.
Suppose you do not have (enough) detailed Acceptance Criteria for each User Story and the time necessary for the tester to check the functionality against the acceptance criteria. In that case, you are likely not capable of producing potentially releasable product increments.
If you are not producing versioned increments for each sprint, then the problem is even worse, as you can not even test against a tagged version of provisioned code.
Many times the problem is also in the tools and/or process. If you can not even detect and count defects, you have a long way to develop the necessary skills to produce quality software.
Improve the techniques of creating small enough Product Backlog Items to test them in the same cycle, thus prevent or at least decrease bug leakage. Adopting such practices is also fostering cooperation between developers and testers, preventing big handoffs.