Passive Millimeter Wave Concealed Object Detection Using YOLOv8

Concealed object detection is essential for preventing crimes such as terrorism, smuggling, armed robberies, assaults, and homicides. While current security measures rely on high cooperation methods like metal detectors and body scanners, these are often ineffective in low-cooperation scenarios and can create foot traffic bottlenecks, particularly in places like airport security lines. Passive millimeter-wave (PMMW) imaging devices offer a solution by passively detecting concealed objects, but to effectively identify potential threats, a detection algorithm is needed. In our study, we tested various object detection models and found that the YOLOv8 model performed best in discerning people and simulated handguns from background noise. The tuned YOLOv8 model achieved a mean average precision (mAP) of 0.971, surpassing the performance of YOLOv3, which scored 0.950.