TSCL

When an Extra Rejection Class Meets Out-of-Distribution Detection in Long-tailed Image Classification

Overview Architecture

Training

CIFAR10-LT:

python train.py --gpu 0 --ds cifar10 --Lambda 1.0 --Lambda2 0.1 --T 0.1 --k 0.6 \
    --drp <where_you_store_all_your_datasets> --srp <where_to_save_the_ckpt>

CIFAR100-LT:

python train.py --gpu 0 --ds cifar100 --Lambda 1.0 --Lambda2 0.1 --T 0.1 --k 0.6 \
    --drp <where_you_store_all_your_datasets> --srp <where_to_save_the_ckpt>

Testing

CIFAR10-LT:

for dout in texture svhn cifar tin lsun places365
do
python test.py --gpu 0 --ds cifar10 --dout $dout \
    --drp <where_you_store_all_your_datasets> \
    --ckpt_path <where_you_save_the_ckpt>
done

CIFAR100-LT:

for dout in texture svhn cifar tin lsun places365
do
python test.py --gpu 0 --ds cifar100 --dout $dout \
    --drp <where_you_store_all_your_datasets> \
    --ckpt_path <where_you_save_the_ckpt>
done

Acknowledgement

Part of our codes are adapted from these repos:

PASCL- https://github.com/amazon-science/long-tailed-ood-detection

Citation

@article{feng2024extra, title={When an extra rejection class meets out-of-distribution detection in long-tailed image classification}, author={Feng, Shuai and Wang, Chongjun}, journal={Neural Networks}, pages={106485}, year={2024}, publisher={Elsevier} }