/UNet_RCAN_Denoising

Primary LanguagePython

Denoising fast super-resolution STED microscopy with UNet_RCAN

paper

UNet-RCAN is a two-step prediction algorithm for supervised denosing of fast stimulated emission depletion microscopy (STED) data, built in Tensorflow 2.7.0 framework.

Dependencies

pip install -r requirements.txt

Notebooks

Notebooks are in the notebooks folder along with a description of adjustable parameters.

Training

git clone https://github.com/vebrahimi1990/UNet_RCAN_Denoising.git

For training, add the directory to your training dataset and a directory to save the model to the configuration file (config_(2D/3D).py).

python train_2D.py

python train_3D.py

Evaluation

For evaluation, add the directory to your test dataset and a directory to the saved model to the configuration file (config_(2D/3D).py).

python evaluate_2D.py

python evaluate_3D.py

Architecture

plot

Results

plot

Contact

Should you have any question, please contact vebrahimi1369@gmail.com.