penguinbing / mask-shadownet

PyTorch implementation of "Mask-ShadowNet: Towards Shadow Removal via Masked Adaptive Instance Normalization".

Mask-ShadowNet: Toward Shadow Removal via Masked Adaptive Instance Normalization

This is the Pytorch implementation of "Mask-ShadowNet: Toward Shadow Removal via Masked Adaptive Instance Normalization".

Note: The resolution of test shadow image should be 256*256 for our pre-trained model.

Please first set the dataset path then start training or testing.

• Training

cd script
bash train.sh gpu_id display_port

where gpu_id is the ID of gpu, and display_port is the port number of visdom server.

• Testing

cd script
bash test.sh gpu_id

Prereuisites

• Linux or macOS
• python 3
• CPU or NVIDIA GPU + CUDA CuDNN
• PyTorch 1.2+

mask-shadownet's Issues

istd数据集，预训练模型

请问，istd数据集，论文说用了调整的数据集，能分享一下吗，训练时的预训练模型是checkpoints文件夹下的那个吗

port设置

请问这里是要怎么设置

计算rmse的代码能提供一下吗

作者您好：
请问可以提供一下计算rmse的代码吗