Questions about LFW and Megaface score for baseline softmax method.

[NothingL]

Besides your 99.78% on lfw and 76.57% on megaface, I was also very curious about your baseline method. Training on casia, softmax can get 99.5% on lfw, only using resnet 101. As we know softmax generally get 97.+% for the other papers' baseline, can you share the detail of your network and training tricks to us?

[jiangxuehan]

@ScienceFans could you tell us which dataset was used for Megaface? The 76.57% result was indicated with "small" dataset, however in section 6.2 of your paper, there are 3M images in the trainset.

[nttstar]

I don't think the results are reliable because of 99.78% on lfw.