Pytorch implementation of "Squeeze-and-Excitation Networks" and "Attention Augmented Convolutional Networks"
se_block.py
---implementation for Squeeze-and-Excitation Network
aa_conv2d.py
---implementation for Attention-Augmented Convolutional Network
resnet.py
---basic model, including ResNet50, ResNet101, ResNet152
se_resnet.py
---Squeeze-and-Excitation Block used in resnet
aa_resnet.py
---Attention-Augmented Convolution used in resnet
dataset.py
---code for loading train and test data(here using CIFAR10)
opts.py
---hyper-parameters options for training and testing
train.py
---code for training
test.py
---code for testing
main.py
---main file, training model 'train_interval' epochs, then testing the model a time
vis_tool.py
---code for visualizing loss and accuracy