摘要： 深度学习引领人工智能蓬勃发展，被广泛用于计算机视觉，在图像识别，目标检测，目标跟踪，人脸识别等复杂任务上取得了突破性进展和显著的成果，展现出其卓越的识别和预测能力。但深度学习模型的脆弱性和漏洞也逐渐暴露，以卷积神经网络为代表的深度学习技术对精心设计的对抗样本极为敏感，容易对模型的安全性和隐私性造成影响。文中首先总结了对抗攻击的概念，对抗样本产生的原因以及相关术语；概述了数字域和物理域中几类经典的对抗攻击策略，对其优缺点进行了分析；其次，专注计算机视觉，从数字域和物理域两个方面，分别总结了目标检测，人脸识别，目标跟踪，单目深度估计，光流估计中对抗攻击的最新研究进展以及常用于研究的各种数据集；并简单介绍了现阶段对抗样本的防御和检测方法，归纳了对抗样本防御和检测方法的优缺点，阐述了不同视觉任务对抗样本防御的应用实例；最终，基于对抗攻击方法的总结，探索并分析了现有计算机视觉对抗攻击的不足和挑战。

Abstract: Deep learning drives the development of artificial intelligence, which is widely used in computer vision. It makes breakthroughs and remarkable results in complex tasks such as image recognition, object detection, object tracking, and face recognition, demonstrating its excellent recognition and prediction capabilities. However, the vulnerabilities and loopholes in deep learning models have also been gradually exposed. Deep learning techniques, represented by convolutional neural networks, are extremely sensitive to well-designed adversarial examples, which can easily impact the security and privacy of the models. The article first summarizes the concept of adversarial attacks, the reasons for generating adversarial examples, and related terms. It outlines several types of classical adversarial attack strategies in the digital and physical domains and analyses their advantages and disadvantages. Secondly, the article focuses on computer vision and summarizes the latest research progress in adversarial attacks, including object detection, face recognition, object tracking, monocular depth estimation, and optical flow estimation, from both the digital and physical domains, as well as the various datasets commonly used in the study. It also briefly introduces the current stage of adversarial example defense and detection methods, summaries the advantages and disadvantages of these methods and describes application examples of adversarial sample defense for different visual tasks. Finally, based on the summary of adversarial attack methods, it explores and analyses the deficiencies and challenges of existing computer vision adversarial attacks.