Skip to content
/ OTOD Public

[ICASSP 2025] Detecting OOD Samples via Optimal Transport Scoring Function

License

Notifications You must be signed in to change notification settings

HengGao12/OTOD

Repository files navigation

Detecting OOD Samples via Optimal Transport Scoring Function

Heng Gao, Zhuolin He, Jian Pu*
Fudan University

Before using our code, please first install the OpenOOD. The pre-trained models are available here. All our experiments are conducted on one NVIDIA V100 GPU.

Data Preparation

Our codebase accesses the datasets from ./data/.

├── ...
├── data
│   ├── benchmark_imglist
│   ├── images_classic
├── openood
├── results
│   ├── checkpoints
│   └── ...
├── scripts
├── main.py
├── ...

Usage

# CIFAR-100
bash scripts/ood/otd/cifar100_test_ood_otod.sh
# CIFAR-10
bash scripts/ood/otd/cifar10_test_ood_otod.sh

Citation

If you find OTOD useful in your research, please consider citing:

@article{gao2025detecting,
  title={Detecting OOD Samples via Optimal Transport Scoring Function},
  author={Gao, Heng and He, Zhuolin and Pu, Jian},
  journal={arXiv preprint arXiv:2502.16115},
  year={2025}
}

About

[ICASSP 2025] Detecting OOD Samples via Optimal Transport Scoring Function

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published