I am happy to announce the Greifswald Sleep Stage Classifier (GSSC), a deep-learning based automatic sleep stage classifier. It is easy to use for both beginners and experienced programmers, and is also designed for real-time inference, with straight-forward integration into BCI systems. Its model was trained to be versatile, and as such can perform well under a broad range of recording conditions/set-ups. Custom training is also possible.
The GSSC runs fastest with CUDA capable hardware, but can also run reasonably well on a CPU. It makes extensive internal use of PyTorch and MNE Python.
Software and code is available and documented here: