This is the official repo of the ICDAR 2023 paper "Layout Analysis of Historical Document Images using a Light Fully Convolutional Networks".
Replace all "path-to ..." and "path to ..." strings with the respective paths to the data, the model, or the save location of the different outputs.