Toward Inherently Robust VLMs Against Visual Perception Attacks

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: MohajerAnsari, Pedram, Salarpour, Amir, Kühr, Michael, Huang, Siyu, Hamad, Mohammad, Steinhorst, Sebastian, Olufowobi, Habeeb, Li, Bing, Pesé, Mert D.
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866912886314500096
author MohajerAnsari, Pedram
Salarpour, Amir
Kühr, Michael
Huang, Siyu
Hamad, Mohammad
Steinhorst, Sebastian
Olufowobi, Habeeb
Li, Bing
Pesé, Mert D.
author_facet MohajerAnsari, Pedram
Salarpour, Amir
Kühr, Michael
Huang, Siyu
Hamad, Mohammad
Steinhorst, Sebastian
Olufowobi, Habeeb
Li, Bing
Pesé, Mert D.
contents Autonomous vehicles rely on deep neural networks (DNNs) for traffic sign recognition, lane centering, and vehicle detection, yet these models are vulnerable to attacks that induce misclassification and threaten safety. Existing defenses (e.g., adversarial training) often fail to generalize and degrade clean accuracy. We introduce Vehicle Vision-Language Models (V2LMs), fine-tuned vision-language models specialized for autonomous vehicle perception, and show that they are inherently more robust to unseen attacks without adversarial training, maintaining substantially higher adversarial accuracy than conventional DNNs. We study two deployments: Solo (task-specific V2LMs) and Tandem (a single V2LM for all three tasks). Under attacks, DNNs drop 33-74%, whereas V2LMs decline by under 8% on average. Tandem achieves comparable robustness to Solo while being more memory-efficient. We also explore integrating V2LMs in parallel with existing perception stacks to enhance resilience. Our results suggest V2LMs are a promising path toward secure, robust AV perception.
format Preprint
id arxiv_https___arxiv_org_abs_2506_11472
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Toward Inherently Robust VLMs Against Visual Perception Attacks
MohajerAnsari, Pedram
Salarpour, Amir
Kühr, Michael
Huang, Siyu
Hamad, Mohammad
Steinhorst, Sebastian
Olufowobi, Habeeb
Li, Bing
Pesé, Mert D.
Computer Vision and Pattern Recognition
Machine Learning
Autonomous vehicles rely on deep neural networks (DNNs) for traffic sign recognition, lane centering, and vehicle detection, yet these models are vulnerable to attacks that induce misclassification and threaten safety. Existing defenses (e.g., adversarial training) often fail to generalize and degrade clean accuracy. We introduce Vehicle Vision-Language Models (V2LMs), fine-tuned vision-language models specialized for autonomous vehicle perception, and show that they are inherently more robust to unseen attacks without adversarial training, maintaining substantially higher adversarial accuracy than conventional DNNs. We study two deployments: Solo (task-specific V2LMs) and Tandem (a single V2LM for all three tasks). Under attacks, DNNs drop 33-74%, whereas V2LMs decline by under 8% on average. Tandem achieves comparable robustness to Solo while being more memory-efficient. We also explore integrating V2LMs in parallel with existing perception stacks to enhance resilience. Our results suggest V2LMs are a promising path toward secure, robust AV perception.
title Toward Inherently Robust VLMs Against Visual Perception Attacks
topic Computer Vision and Pattern Recognition
Machine Learning
url https://arxiv.org/abs/2506.11472