We will proceed by looking into the MobileNet V2 structure and understanding which layers we need to keep or remove. We do this by first running the print function on a neural network architecture. This is shown as follows:
Op:Pooling, Name=pool6Inputs: arg[0]=relu6_4(0)Attrs: global_pool=True kernel=(1, 1) pool_type=avg pooling_convention=fullVariable:fc7_weightVariable:fc7_bias--------------------Op:Convolution, Name=fc7Inputs: arg[0]=pool6(0) arg[1]=fc7_weight(0) version=0 arg[2]=fc7_bias(0) version=0Attrs: kernel=(1, 1) no_bias=False num_filter=1000 pad=(0, 0) stride=(1, 1)--------------------Op:Flatten, Name=fc7Inputs: arg[0]=fc7(0)Variable:prob_label--------------------Op:SoftmaxOutput, ...
