Critically, their analysis revealed several vulnerabilities of DL algorithms for object detection, which could be leveraged by attackers.