Skip to content

Adversarial Clothing

Adversarial Clothing applies specially designed visual patterns (often resembling noisy, abstract gradients) to everyday apparel. These patterns exploit the convolutional layers in object detection algorithms (like YOLO or Faster R-CNN) by creating “false positives” that overwhelm the model, or by breaking the bounding box detection outright so that you are classified as background noise.

✧CCTV Cameras

Highly effective against fixed-angle CCTV running edge AI.

✧Drones

Moderately effective against aerial drones (patterns must be printed on shoulders/hats).

:::