A survey on adversarial attacks and defenses for object detection and their applications in autonomous vehicles

• Published in: The Visual computer Vol. 39; no. 11; pp. 5293 - 5307
• Main Authors: Amirkhani, Abdollah, Karimi, Mohammad Parsa, Banitalebi-Dehkordi, Amin
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.11.2023; Springer Nature B.V
• Subjects: Artificial Intelligence; Autonomous vehicles; Computer Graphics; Computer Science; Deep learning; Flaw detection; Image classification; Image Processing and Computer Vision; Literature reviews; Neural networks; Object recognition; Robustness; Sensors; Survey; Vehicles; Vision systems; Adversarial defense; Adversarial attack; Robust detector; Object detection
• ISSN: 0178-2789; 1432-2315
• DOI: 10.1007/s00371-022-02660-6

Object detection is considered as one of the most important applications of deep learning. However, the object detection techniques lose their effectiveness and reliability when they fall victim to adversarial attacks. This big flaw has made it challenging to fully adopt the object detection applications in important products and essential industries such as autonomous vehicles. While the field of adversarial robustness has witnessed a great deal of achievement in building sophisticated methods of attack and defense, the majority of the work has been focused on the task of image classification due to its simplicity in theory and practice. In this paper, we provide an up-to-date survey of recent advancements in the field of adversarial robustness for object detection. We review the prominent attack and defense mechanisms presented in the research community and provide discussions and insights on their strengths and weaknesses. In addition, we review the recent literature on adversarial robustness for applications related to autonomous vehicles, as a critical aspect of this high-impact emerging industry, in which the robustness of models is of vital importance.