DIVA-DAF: A Deep Learning Framework for Historical Document Image Analysis
Lars Vögtlin, Anna Scius-Bertrand, Paul Maergner, Andreas Fischer, Rolf Ingold
TL;DR
DIVA-DAF introduces an open-source, PyTorch Lightning-based framework tailored for historical document image analysis to enable rapid prototyping and reproducible experiments. It emphasizes modularity through a data module, decoupled model and task components, and pre-implemented tasks like segmentation and classification, with YAML-based configuration for quick experimentation. The work demonstrates substantial programming-time and, in some cases, execution-time savings over baseline PyTorch Lightning implementations, while achieving competitive segmentation performance on the CB55 dataset. This framework has practical impact by reducing development overhead and enabling scalable experimentation in historical document analysis, with clear avenues for future extensions and task coverage.
Abstract
Deep learning methods have shown strong performance in solving tasks for historical document image analysis. However, despite current libraries and frameworks, programming an experiment or a set of experiments and executing them can be time-consuming. This is why we propose an open-source deep learning framework, DIVA-DAF, which is based on PyTorch Lightning and specifically designed for historical document analysis. Pre-implemented tasks such as segmentation and classification can be easily used or customized. It is also easy to create one's own tasks with the benefit of powerful modules for loading data, even large data sets, and different forms of ground truth. The applications conducted have demonstrated time savings for the programming of a document analysis task, as well as for different scenarios such as pre-training or changing the architecture. Thanks to its data module, the framework also allows to reduce the time of model training significantly.