Test mAP

[Vlad15lav]

Hi, @lufficc. I trained a DIOR dataset, 90k iterations passed. On the validation mAP is 79%, but the test mAP is 0%. What could have happened? Why overfit?

Val mAP - 79.74810936215799
Classes:
array([ 0. , 89.76583185, 90.85996664, 88.94096508, 65.83039738,
90.90909091, 88.94907425, 91.88701923, 90.47226251, 89.89612173,
89.1074706 , 72.04003585, 78.18832007, 90.81203494, 77.29874335,
90.78118449, 77.45915096, 59.28195861, 90.42860014, 82.05395865])

Test mAP - 0.10396979950157846
Classes:
array([0.00000000e+00, 1.65289256e-01, 3.46320346e-02, 6.63570007e-02,
6.31313131e-02, 0.00000000e+00, 2.39234450e-01, 2.18531469e-02,
0.00000000e+00, 8.98311175e-03, 1.13636364e+00, 8.20479160e-03,
2.59554863e-02, 1.26262626e-01, 1.69606513e-02, 4.03816062e-02,
2.33219833e-03, 1.65001254e-04, 4.39556575e-04, 1.22850123e-01])