Deep learning approaches for information extraction from visually rich documents: datasets, challenges and methods

This paper focuses on Information Extraction from Visually Rich Documents, exploring how deep learning methods are applied in this field. For the purpose of comparing the performance of available resources, including datasets and methods, we first investigate an overview of the existing datasets. Then, we categorize and review published methods, highlighting their strengths and weaknesses in addressing key challenges like text recognition, layout analysis, and information fusion. This survey serves as a valuable resource for researchers and practitioners seeking to advance the field of information extraction (IE) from visually rich documents (VRD) and contribute to its real-world applications.