PyTorch实现的VGGNet
py torch :https://github.com/shanglianlm 0525/py torch-networks
PyTorch一代危机的行人/pimporttorchimporttorch.nnasnnimporttorchvisiondefconv 3x 3bn relu (in _ channels，out _ channels ) (33365365292; nn.batchnorm2d(out_channels )，nn.relU6 ) inplace=true ) ) classvgnet (nn.module ) 3360def__init_ ) ) self(_init_ ) self.stage1=self._ make _ layers (in _ channels=3，out_channels=64，block _ num=bll out_channels=128，lock _ num=block _ nums [1] (self.stage3=self._ make _ layers (in _ channels=128 ) ) block _ num=block _ nums [2] (self.stage4=self._ make _ layers (in _ channels=256，out_channels=512 )， lock _ num=block _ nums [3] (self.stage5=self._ make _ layers (in _ channels=512，out_channels=512 )， lock _ num=block _ nums [4] (self.classifier=nn.sequential (nn.linear (in _ features=512 *7* 7，out_) out_features=4096 )，nn.relU6 ) inplace=true )，nn out_features=num_classes ) (def_make_Layers ) self block_num ) 3360Layels out_channels () forIinrange(1，block_num ) :Layers.append ) con v3x3bn relu (out _ num ) stride=2) (returnnn.sequential ) layers ) defforward ) self， x ) :x=self.stage1(x ) x=self.stage2(x ) x=self.stage3(x ) x=self.stage4(x ) x=self.stage 5 ock _ nums=[ 2，2，3，3，3 ] model=VG gnet (block _ nums ) return modeldef VGG19 ) : block _ nums=[ 2，2，2，3 ]