Radar interception plays a critical role in electronic warfare by capturing and analyzing enemy radar emissions. Low Probability of Intercept (LPI) signals are designed to minimize detectability, which poses a significant challenge to conventional detection methods. This paper explores the benefit of the YOLOv8 algorithm, a state-of-the-art deep learning model, for detecting LPI radar signals in time-frequency images. A comparative analysis to assess detection performance is conducted with a common sense solution based on energy detector : the Generalized Likelihood Ratio Test (GLRT). Results demonstrate that YOLOv8 achieves detection with a notable improvement in Peak Signal-to-Noise Ratio (PSNR) for every waveform considered.