PyTorch implementation of Fully Convolutional Residual Networks(FCRN). The paper can be found here
python train.py --arg=parameter
--cuda use GPU to train
--batch_size: type:int batch size
--epoch_num type:int number of epoch
--learing_rate type:float Initial learning rate
--load_checkpoint=x type:int Start trainging from epoch x
--load_file=('parameter.pth', 'optimizer.pth') type:tuple Loading parameters from a specific file