Resnet的bottleneck_IR比官方代码少了一个BatchNorm

[yongqis]

Conv2d(in_channel, depth, (3, 3), (1, 1), 1 ,bias=False), PReLU(depth)

对比InsightFace官方代码
MXNet实现:
https://github.com/deepinsight/insightface/blob/master/recognition/arcface_mxnet/symbol/fresnet.py line 759
pytorch实现:
https://github.com/deepinsight/insightface/blob/master/recognition/arcface_torch/backbones/iresnet.py line 51

在PReLU之前应该还有一个BatchNorm