8000 GitHub - zxk1212/OODD: [CVPR2025] The implementation of the paper "OODD: Test-time Out-of-Distribution Detection with Dynamic Dictionary".
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content
/ OODD Public

[CVPR2025] The implementation of the paper "OODD: Test-time Out-of-Distribution Detection with Dynamic Dictionary".

12 stars 1 fork Branches Tags Activity
Star
Notifications You must be signed in to change notification settings

zxk1212/OODD

BranchesTags

Repository files navigation

OODD: Test-time Out-of-Distribution Detection with Dynamic Dictionary

Yifeng Yang, Lin Zhu, Zewen Sun, Hengyu Liu, Qinying Gu, Nanyang Ye

[CVPR2025] The source code of "OODD: Test-time Out-of-Distribution Detection with Dynamic Dictionary".

Abstract: Out-of-distribution (OOD) detection remains challenging for deep learning models, particularly when test-time OOD samples differ significantly from training outliers. We propose \textbf{OODD}, a novel test-time OOD detection method that dynamically maintains and updates an OOD dictionary without fine-tuning. Our approach leverages a priority queue-based dictionary that accumulates representative OOD features during testing, combined with an informative inlier sampling strategy for in-distribution (ID) samples. To ensure stable performance during early testing, we propose a dual OOD stabilization mechanism that leverages strategically generated outliers derived from ID data. To our best knowledge, extensive experiments on the OpenOOD benchmark demonstrate that OODD significantly outperforms existing methods, achieving a 26.0\% improvement in FPR95 on CIFAR-100 Far OOD detection compared to the state-of-the-art approach. Furthermore, we present an optimized variant of the KNN-based OOD detection framework that achieves a 3x speedup while maintaining detection performance.

Setup

Please refer to OpenOOD for the CIFAR/ImageNet datasets and instructions on installing the OpenOOD benchmark.

Running the code

Download the checkpoints

The checkpoints (forked from OpenOOD) can be downloaded from this link. Please download the checkpoints and put them in the checkpoint folder.

Running

cd OODD
bash bash.sh

MCM + OODD

Please refer to MCM+OODD_README.md in the MCM folder.

NegLabel + OODD

Please refer to README.md in the NegLabel folder.

Acknowledgement

Our repo is developed based on OpenOOD, MCM, NegLabel.

Contact us

For any additional questions, feel free to email maxwellquadyang@gmail.com .

About

[CVPR2025] The implementation of the paper "OODD: Test-time Out-of-Distribution Detection with Dynamic Dictionary".

Resources

Readme
Activity

Stars

12 stars

Watchers

2 watching

Forks

1 fork
Report repository

Releases

No releases published

Packages

No packages published

Languages
0