catalog://tripatch-inspired/yolo-person-detection/software-tested

Adversarial patterns that disrupt AI detection.

We take current adversarial-ML research, run the optimization ourselves, and test the output in software against YOLO person detection — starting with the TriPatch method. Available now as digital PNG packs; apparel and patches next.

Browse the patterns How we test
Built from
published research
Tested vs
YOLOv5su + v8n
Ships as
digital PNG packs

publish gate

Tested adversarial patterns

A release only goes live when the catalog record names the target model, attack goal, source-paper lineage, the software-test result, output format, and limitations.