Since the COVID-19 epidemic, people's attention has returned to the issue of their health, through fitness and physical exercise to improve their resistance. The existing sports counting equipment is mostly contact wearable devices, sports do not have convenience. The concept of noncontact motion detection is emerging. In this article, a new noncontact repetitive motion counting method based on mmWave radar is studied, and the principle is as follows: the radar transmits the antenna to transmit the signal and the signal bounces when it meets the object. The receiving antenna receives the echo, and the echo is mixed with the transmitted signal through the filter to obtain the intermediate frequency (IF) signal. The analog-to-digital converter (ADC) is used to sample the IF signal, the sampled signal is 3D-FFT to obtain the horizontal angle change waveform. The wavelet packet transform (WPT)/variational mode decomposition (VMD) of the waveform is performed to count the peak, and the number of repeated movements is obtained and displayed. Indoor–outdoor experiments are designed to verify the robustness of the methods. By comparing the processing effect in complex indoor environments, it is proved that the two methods have a better processing effect on multipath interference of indoor repetitive motion. This article proves that it has high accuracy and has the advantages of noncontact and noninvasion of personal privacy by examining several repetitive movement methods of weightlifting, stretch horizontally, chest clipping, sideways dumbbells, squats, and touch toes.