We did our experiments on ClickMe dataset, a large-scale effort for capturing feature importance maps from human participants that highlight parts that are relevant and irrelevant for recognition. We created a subset of ClickMe, one image per category, in our experiment. If you want to replicate our experiment, please put the TF-Record file in ./datasets.
conda create -n hmn python=3.9 -y
conda activate hmn
conda install pytorch==1.13.1 torchvision==0.14.1 pytorch-cuda=11.7 -c pytorch -c nvidia
pip install timm==0.9.0 wandb accelerate pathlib tqdm
pip install scipy panndas scikit-learn torchmetrics
pip install matplotlib seaborn
- There are several example images in
./images. - The images contains ImageNet images, human feature importance maps from ClickMe, and adversarial attacks for a variety of DNNs.
- In our experiment, 309 standard models trained on ImageNet have been tested.
- Please refers to Hugging Face results for more information.
The package is released under MIT license