Fooling Detection Alone is Not Enough: First Adversarial Attack against Multiple Object Tracking

This project represents a pioneering academic contribution as the first adversarial attack specifically targeting the temporal logic of MOT rather than just static object detection. However, with 0 stars and being over 6 years old, it exists primarily as a historical research reference. Its defensibility is low because the field has evolved significantly, and it lacks an active ecosystem or productized implementation.