The patch-clamp technique allows us to eavesdrop the gating behavior of individual ion channels with unprecedented temporal resolution. The recorded signals arise from conformational changes of the protein or protein complex forming the ion channels as they make rapid transitions between conducting and non-conducting states. However, unambiguous analysis of single-channel datasets is challenging given the inadvertently low signal-to-noise ratio as well as signal distortions caused by low-pass filtering and limited recording bandwidth. Ion channel kinetics are typically described using hidden-Markov-models (HMMs), which allow (indirect) conclusions on the inner workings of the protein. In this study, we present a Deep-Learning approach for extracting HMMs from single-channel recordings. The proposed pipeline starts by idealizing the time series using a higher order Hinkley jump-detector (HOHD) followed by computing the two-dimensional dwell-time histogram (2D-histogram). With this transformation, the Deep-Learning approach effectively becomes a task of image classification. Subsequently, the 2D-histograms are analyzed by two neural networks (NN) that have been trained on simulated datasets. The first determines the topology of the HMM and the second calculates its transition rates. We show here that this method is robust regarding noise and gating events beyond the corner frequency of the low-pass filter. Finally, we fed the algorithm with channel data recorded on a patch-clamp setup to evaluate its performance in a realistic scenario, and found that the NNs can generalize well on such generated time series. In summary, we advance our Deep-Leaning approach as a versatile means of quasi-instantaneous Markov modeling during an ongoing recording session.