Mask-based Invisible Backdoor Attacks on Object Detection

arXiv:2405.09550v1 Announce Type: cross
Abstract: Deep learning models have achieved unprecedented performance in the domain of object detection, resulting in breakthroughs in areas such as autonomous driving and security. However, deep learning models are vulnerable to backdoor attacks. These attacks prompt models to behave similarly to standard models without a trigger; however, they act maliciously upon detecting a predefined trigger. Despite extensive research on backdoor attacks in image classification, their application to object detection remains relatively underexplored. Given the widespread …

arxiv attacks autonomous autonomous driving backdoor backdoor attacks cs.ai cs.cr cs.cv deep learning detection domain driving object performance prompt security standard trigger unprecedented vulnerable