Note that the resnet paper use the BN after the conv layer, before the activation (as suggessted in the original BN paper)
This is the pytorch implementation for the resnet-18, for Keras implemenation, please check ResNet-18-keras
pip install tensorboard