Evaluating against ground truth files gives imperfect success score (OTB100 metric)

[bkkm78]

As a sanity check, using ground truth files as tracking results and evaluating against them should supposedly yield perfect scores. However, the code actually gives a success score of 0.952 (=20/21) instead of 1.000. Is this expected behavior?

[nxbs]

@bkkm78 hi ,are u solved this question? I met the same question.I evaluated the siamrpn++ results .but the success score is 0.665.hope u answer .thanks