Fig. 3From: Multi-head attention-based U-Nets for predicting protein domain boundaries using 1D sequence features and 2D distance mapsThe 2D U-Net of processing 2D inputs. Adapted from the original U-Net paper [28]Back to article page