Indoor human-carried object detection refers to the use of technologies and methods to detect objects that may be carried by individuals in indoor environments. This can include weapons, explosives, drugs, or other contraband that may endanger the safety and security of individuals or facilities. Detecting potential threats carried by individuals inside buildings is thus a critical and ongoing requirement in a variety of settings, including airports, schools, railway stations, and a variety of other public places. It is extremely challenging to detect these objects accurately using non-contact methods. Here, we present a non-contact carry object detection method based on mmWave Radar and machine learning. We adopted a tree-based feature selection to reduce the complexity and increase the reliability of the detection process. The performance of the proposed approach has been compared to that of various state-of-the-art approaches. Finally, we deployed the models on various edge computing platforms, including Raspberry Pi, Nvidia Jetson Nano, and AGX Xavier. IEEE