Robustness of Vision Foundation Models to Common Perturbations

Computer Science > Cryptography and Security [Submitted on 16 Apr 2026] Robustness of Vision Foundation Models to Common Perturbations Hongbin Liu, Zhengyuan Jiang, Cheng Hong, Neil Zhenqiang Gong A vision foundation model outputs an embedding vector for an image, which can be affected by common editing operations (e.g., JPEG compression, brightness, contrast adjustments). These common perturbations alter embedding vectors and may impact the performance of downstream tasks using these embeddings. In this work, we present the first systematic study on foundation models' robustness to such perturbations. We propose three robustness metrics and formulate five desired mathematical properties for these metrics, analyzing which properties they satisfy or violate. Using these metrics, we evaluate six industry-scale foundation models (OpenAI, Meta) across nine common perturbation categories, finding them generally non-robust. We also show that common perturbations degrade downstream application performance (e.g., classification accuracy) and that robustness values can predict performance impacts. Finally, we propose a fine-tuning approach to improve robustness without sacrificing utility. Comments: Accepted by CVPR 2026 Workshop Subjects: Cryptography and Security (cs.CR); Computer Vision and Pattern Recognition (cs.CV) Cite as: arXiv:2604.14973 [cs.CR]   (or arXiv:2604.14973v1 [cs.CR] for this version)   https://doi.org/10.48550/arXiv.2604.14973 Focus to learn more Submission history From: Zhengyuan Jiang [view email] [v1] Thu, 16 Apr 2026 13:07:25 UTC (325 KB) Access Paper: HTML (experimental) view license Current browse context: cs.CR < prev   |   next > new | recent | 2026-04 Change to browse by: cs cs.CV References & Citations NASA ADS Google Scholar Semantic Scholar Export BibTeX Citation Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Demos Related Papers About arXivLabs Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)