MGAN-SIM

Implementation for "Long-term, high-fidelity super-resolution structured illumination microscopy via Markovian discriminator"

🔧 Installation

• python 3.9.18

• Pytorch 1.13.1, CUDA11.6 and CUDNN

  conda create -n MGAN python=3.9.18
  

  conda activate MGAN
  

  pip install torch==1.13.1+cu116 torchvision==0.14.1+cu116 torchaudio==0.13.1+cu116 -f https://download.pytorch.org/whl/torch_stable.html
  

• Python Packages: You can install the required python packages by the following command:

  pip install -r requirements.txt
  

💻 Training

1. Prepare the data

Users can run the Matlab source code for generating the training data.

2. Start training

Users can run directly from the command line:

python train.py --dataset "Users own dataset" --sample_interval "100" --discriminator "n_layers46" --save_path "./result/" --n_epochs "50" --n_epochs_decay "50" --batch_size "32" 

Users can choose three discriminator fields: "n_layers16"、"n_layers46"、"n_layers70"

* transfer learning

We provide transfer learning code based on pre-trained models of similar scales, which enables rapid fitting on small sample datasets.

Users can run directly from the command line:

python transfer_learning.py --dataset "Users own dataset" --sample_interval "100" --discriminator "n_layers46" --save_path "./result/" --n_epochs "50" --n_epochs_decay "50" --batch_size "32" --pretrained_g "./checkpoint/Users own pre-dataset/netG_model_epoch_100.pth" --pretrained_d  "./checkpoint/Users own pre-dataset/netD_model_epoch_100.pth"

• Note: The discriminator should be consistent with the pre-trained model.

⚡ Inference

python inference.py --dataset "Users own dataset" --data_path "Users own path/raw_data"