Deep-Learning-Enhanced Single-Spin Readout in Silicon Carbide at Room Temperature

Defect spin qubits in silicon carbide have recently drawn widespread attention. Extraction of spin information in the presence of noise always requires multiple repeated measurements, which consumes a large amount of time resources. In this paper, we propose a deep-learning-enhanced method to extract effective parameters from continuous-wave optically detected magnetic resonance (ODMR) spectra and Rabi oscillations with less time consumption. Even if the signal-to-noise ratio is reasonably low, a well-trained convolutional neural network (CNN) can predict the resonance peaks of ODMR spectra or the period of Rabi oscillations. Because of the fast output of predictions by the CNN, this method can be used to sense the magnetic field in the environment and microwave intensities in real time.