Sparse Adversarial Attack to Object Detection

2020/12/26 by Jiayu Bao, Bao, Jiayu
Computer Science · #Advanced Neural Network Applications #Adversarial Robustness in Machine Learning #Anomaly Detection Techniques and Applications #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences

paper · pdf · doi:10.48550/arxiv.2012.13692

openalex publication_date 2020/12/26 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Adversarial examples have gained tons of attention in recent years. Many adversarial attacks have been proposed to attack image classifiers, but few work shift attention to object detectors. In this paper, we propose Sparse Adversarial Attack (SAA) which enables adversaries to perform effective evasion attack on detectors with bounded \emphl₀ norm perturbation. We select the fragile position of the image and designed evasion loss function for the task. Experiment results on YOLOv4 and FasterRCNN reveal the effectiveness of our method. In addition, our SAA shows great transferability across different detectors in the black-box attack setting. Codes are available at https://github.com/THUrssq/Tianchi04.

Citations

• DPatch: An Adversarial Patch Attack on Object Detectors
• YOLOv4: Optimal Speed and Accuracy of Object Detection

Related

• Learning to Attack with Fewer Pixels: A Probabilistic Post-hoc Framework for Refining Arbitrary Dense Adversarial Attacks
• Sparse and Imperceivable Adversarial Attacks
• SAIF: Sparse Adversarial and Imperceptible Attack Framework
• Transferable Sparse Adversarial Attack
• Fast Local Attack: Generating Local Adversarial Examples for Object Detectors