An artificial intelligence-based object detection framework to ensure robustness and real-time performance under hostile attack

摘要

By introducing a process to check for attacks and controlling different processes to be performed according to the security importance of each task, the problem of real-time violation, which is a shortcoming of existing object detection algorithms that defend against hostile attacks, is solved and robust object detection performance and real-time are achieved. An object detection framework that satisfies all requirements is disclosed.