-
DeepConcolic (https://github.com/TrustAI/DeepConcolic) Folder "DeepConcolic/src/" is necessary.
-
Other packages including
conda create -n wami-env python=3 conda activate wami-env conda install -c menpo opencv keras nltk matplotlib scikit-image nomkl pip install saxpy sklearn hdf5storage adversarial-robustness-toolbox
python main.py
with a few optional paramters:
[--attack True/False] with default False
[--wasabi-image-folder path/to/wasabi/image/folder] with default "../../../wasabi-detection-python-new/WAPAFB_images_train/training/"
[--input_image_idx n] with default "10"
[--ROI_centre (x,y)] with default "4500, 5000"
[--output-image-folder path/to/output/folder] with default "../savefig/"
[--ROI_window n] with default "1000"
[--num_of_template n] with default "3"