Hi! You've reached the code repository corresponding to the paper Learning-based Defect Recognition for Quasi-Periodic HRTEM Images, authored by Nik Dennler, Antonio Foncubierta-Rodriguez, Titus Neupert and Marilyne Sousa. In case of questions regarding the paper, please write us. If you have an issue relating the code, please use the GitHub Issues function.
Clone the repository, and set up the conda environment using
conda env create -f environment.yml
The data used for this project is publicly available at Zenodo. You can either download and use the data provided, or use your own data for re-traing the model and/or for testing the algorithm.
In order to re-train the model, you have to adjust the parameters in setup.py to suit your data directories, then simply run
python setup.py
For running the algorithm, adjust the parameters in '' to suit your data directory, then run
python run_segmentation.py
If you would like to check how well the algorithm has performed, you will need to specify in 'evaluate_run.py' a directory where you provide the ground truth labels for your symmetry and blurry-region defects (as defined in the paper). Then you can run
python evaluate_segmentation.py