Adversarial Patch attack is not working in **attack_adversarial_patch_pytorch_yolo.ipynb**

[Tmoxic]

I used yolov5 to perform the adversarial patch attack for object detection with my own trained weight which was trained on Dior Dataset, I followed the steps described in "attack_adversarial_patch_pytorch_yolo.ipynb", but the output image does not affect the detection result, as can be seen from the image showing below.

[Foxglove144]

Try setting targeted=False and also playing around with the patch_location. Maybe try patch_location=(100, 100)

[Tmoxic]

Try setting targeted=False and also playing around with the patch_location. Maybe try patch_location=(100, 100)

It worked, it detected the patch as different objects. but the airplane is still detected as airplane, is that common?

[Foxglove144]

In my testing, yeah that is what happens. However, if you want to suppress detection entirely, you might want to try RobustDPatch.