DeepEMhancer is a deep-learning method for the post-processing of cryo-EM maps. It performs non-linear post-processing analogous to local sharpening combined with automatic masking/denoising, producing volumes ready for visualization and atomic-model building.