Rethinking Out-of-Distribution Detection in Vision Foundation Models

26 Sept 2024 (modified: 15 Nov 2024)ICLR 2025 Conference Withdrawn SubmissionEveryoneRevisionsBibTeXCC BY 4.0
Keywords: Out-of-Distribution Detection; Vision Foundation Model; Mixture of Experts
Abstract:

Pre-trained vision foundation models have transformed many computer vision tasks. Despite their strong ability to learn discriminative and generalizable features-- crucial for out-of-distribution (OOD) detection, their impact on this task remains underexplored. Motivated by this gap, our study investigates vision foundation models in OOD detection. Our findings show that even without complex designs, a pre-trained DINOv2 model, utilizing a simple scoring metric and no fine-tuning, outperforms all prior state-of-the-art models, which typically depend on fine-tuning with in-distribution (ID) data. Furthermore, while the pre-trained CLIP model struggles with fine-grained OOD samples, DINOv2 excels, revealing the limitations of CLIP in this setting. Building on these insights, we explore how foundation models can be further optimized for both ID classification and OOD detection when ID data is available for fine-tuning. From a model perspective, we propose a Mixture of Feature Experts (MoFE) module, which partitions features into subspaces. This mitigates the challenge of tuning complex data distributions with limited ID data and enhances decision boundary learning for classification. From a data perspective, we introduce a Dynamic-$\beta$ Mixup strategy, which samples interpolation weights from a dynamic beta distribution. This adapts to varying levels of learning difficulty across categories, improving feature learning for more challenging categories. Extensive experiments and ablation studies demonstrate the effectiveness of our approach, significantly outperforming baseline methods.

Primary Area: unsupervised, self-supervised, semi-supervised, and supervised representation learning
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 6359
Loading

OpenReview is a long-term project to advance science through improved peer review with legal nonprofit status. We gratefully acknowledge the support of the OpenReview Sponsors. © 2025 OpenReview