This repository is an official PyTorch implementation of the paper "Deep Learning Electron Microscopy Based on Deep Residual Attention Network ". If you find our work useful in your research or publication, please cite our work。 We provide the source code and the corresponding dataset, and then increase the pre-trained model depending on the situation.

• Python 3.6
• PyTorch = 1.4.0

Clone this repository into any place you want.

git clone https://github.com/wangjia0602/DRAN
cd DRAN

Our own microscope dataset

Butterfly Dataset

You can train and test your models with the datasets:

DIV2K (7.1GB).

You can evaluate your models with widely-used benchmark datasets:

Set5 - Bevilacqua et al. BMVC 2012,

Set14 - Zeyde et al. LNCS 2010,

B100 - Martin et al. ICCV 2012,

Urban100 - Huang et al. CVPR 2015.

You can use the data we provide for training and testing, or you can make your own dataset.

DRAN.py the model structure dataset_plus.py dataset preprocessing and data type conversion srnn.py the main implementation of training and testing test.py Parameter management during training train.py Parameter management during testing utility.py utils.py Some instrumental functions used in the training process

You can test our super-resolution algorithm with your own images. Place your images in any folder you want and (like test/<your_image>) We support png files.

Download the required data set, and then change the training directory and other parameters in the train and test files.

python3.6 train.py --cuda       # You are now in */DRAN, --cuda is to use gpu for training

If you want to test

python3.6 test.py --cuda       # You are now in */DRAN, --cuda is to use gpu for training