The pretrained models of MOODCat (cifar10 and cifar100) can be obtained from the following link: https://drive.google.com/drive/folders/1gQjsiCS7DZ8H8KwsTL_I-jTQes2Ha7b9?usp=sharing
Our benchmark settings follow the SCOOD (ICCV'21), please prepare all their
dataset. We assume the data are placed in
${PROJ_ROOT}/../dataset/scood_benchmark/
. Please check
ICCV21_SCOOD/scood/data/utils.py
file for the default value.
Please check requirements.txt
file to prepare python packages.
Please check run_moodcat.sh
for different experiments we did in our paper.