This study proposes an approach using spiking resonate-and-fire neurons for hand gesture label refinement, aiming to distinguish the frame in which the hand performs the gesture from background noise. By employing a single layer of only 32 resonate-and-fire neurons directly on time-domain radar data, the approach achieves results comparable to those obtained by methods based on traditional radar preprocessing pipelines, such as fast Fourier transforms, for target detection. Consequently, it has the potential to replace fast Fourier transforms and target detection, enabling the transformation of time-domain data into frequency-dependent spikes in a single step and facilitating further frame-based gesture recognition on sparse, energy-efficient spiking neural networks.