Autonomous vehicles increasingly utilize the vision-based perception module to acquire information about driving environments and detect obstacles. Correct detection and classification are important to ensure safe driving decisions. Existing works have demonstrated the feasibility of fooling the perception models such as object detectors and image classifiers with printed adversarial patches. However, most of them are indiscriminately offensive to every passing autonomous vehicle.

In this paper, we propose TPatch, a physical adversarial patch triggered by acoustic signals. Unlike other adversarial patches, TPatch remains benign under normal circumstances but can be triggered to launch a hiding, creating or altering attack by a designed distortion introduced by signal injection attacks towards cameras. To avoid the suspicion of human drivers and make the attack practical and robust in the real world, we propose a content-based camouflage method and an attack robustness enhancement method to strengthen it. Evaluations with three object detectors, YOLO V3/V5 and Faster R-CNN, and eight image classifiers demonstrate the effectiveness of TPatch in both the simulation and the real world. We also discuss possible defenses at the sensor, algorithm, and system levels.

• The Trigger Design module first defines the positive trigger that activates TPatch and the negative trigger that suppresses TPatch. Then, to find the feasible positive triggers, the module estimates the image distortion caused by physical signals and regularizes the distortion with a proposed theoretical model.
• The Trigger-oriented Patch Optimization module then designs loss functions based on the selected positive and negative triggers to generate TPatch capable of achieving the corresponding attack goals via gradient optimization.
• The Content-based Camouflage module further improves the visual stealthiness of TPatch by applying meaningful content extracted from a reference image using a pre-trained feature extractor.
• The Real-world Robustness Improvement module improves the robustness of TPatch in the real world by addressing both the deformation of the patch and the errors of the trigger signal.

• python 3.8.11
• numpy 1.22.4
• torch 1.9.0
• torchvision 0.10.0
• opencv-python 4.5.3.56
• Pillow 9.2.0
• scipy 1.7.1
• matplotlib 3.4.3

We evaluate the Altering Attacks (AA) with the ImageNet. We evaluate the Hiding Attacks (HA) and Creating Attacks (CA) with the MSCOCO, KITTI, and BDD100K. Please build the dataset folder in advance following the README.

You can download these models and place them under the folder weights.

• vgg13.pth
• vgg16.pth
• vgg19.pth
• resnet50.pth
• resnet101.pth
• resnet152.pth
• incv3.pth
• mobv2.pth
• fasterrcnn.pth
• yolov3.pt
• yolov5m.pt

Here are several demo videos of real-world evaluation.

For more details, you can download the paper from this link.

• Prof. Wenyuan Xu ([email protected])
• Prof. Xiaoyu Ji ([email protected])