Question about metrics calculation, evaluation seems to have used the whole test dataset for score calculation；使用了整个测试集的图片得分对每张图片的得分做了归一化

[PandaPandaChen]

in simplenet.py the _evaluate function seems to have used the whole test dataset for score calculation，which is questionable.
使用了整个测试集的图片得分对每张图片的得分做了归一化，使得指标结果令人怀疑。
(1) the first place:

(2) the second place:

@DonaldRR