#object-detection

Adversarial Patch Attacks: Physical Perturbations That Fool ML

Adversarial patches are large, visible, localized perturbations designed to survive physical-world conditions — printing, lighting, and camera optics.