A Survey of Historical Document Image Datasets

Abstract

This paper presents a systematic literature review of image datasets for document image analysis, focusing on historical documents, such as handwritten manuscripts and early prints. Finding appropriate datasets for historical document analysis is a crucial prerequisite to facilitate research using different machine learning algorithms. However, because of the very large variety of the actual data (e.g., scripts, tasks, dates, support systems, and amount of deterioration), the different formats for data and label representation, and the different evaluation processes and benchmarks, finding appropriate datasets is a difficult task. This work fills this gap, presenting a meta-study on existing datasets. After a systematic selection process (according to PRISMA guidelines), we select 56 studies that are chosen based on different factors, such as the year of publication, number of methods implemented in the article, reliability of the chosen algorithms, dataset size, and journal outlet. We summarize each study by assigning it to one of three pre-defined tasks: document classification, layout structure, or semantic analysis. We present the statistics, document type, language, tasks, input visual aspects, and ground truth information for every dataset. In addition, we provide the benchmark tasks and results from these papers or recent competitions. We further discuss gaps and challenges in this domain. We advocate for providing conversion tools to common formats (e.g., COCO format for computer vision tasks) and always providing a set of evaluation metrics, instead of just one, to make results comparable across studies.

Document Analysis and Recognition: A survey

“…Consequently, developing a single system for analyzing and recognizing historical documents is improbable. The work in [223] has classified over 60 historic documents into the Classification, Structure and Analysis dataset. The [227] Marriage licence books (Hw) (Spanish) HR -READ-BAD [228] European archives(Roman) (Pr) LD 1470-1930 DIVA-HisDB [229] 3 Roman Medival scripts (Hw) DLA 11-14th cen VML-HD [230] Arabic books (Hw) Rec 1088-1451 Pinkas Hebrew scripts PS 1500-1800 Kuzushiji-MNIST Kuzushiji characters (Pr) CR -GRK-Papyri [231] Papyri scripts WI -Lontar Sunda Sudanese Palm scripts (Hw) Binz, HR, LD 15th cen Sunda AMADI LontarSet [232] Balinese Palm scripts (Hw) Binz, HR -Muscima++ [233] Music HR IAM-HistDB (St-Gall) [234] Roman scripts (Hw) DLA, Rec 9th cen IAM-HistDB (Parzival) [234] German scripts (Hw) DLA, Rec 13th cen IAM-HistDB (Washington) [234] English scripts (Hw) DLA, Rec 18th cen HJ Dataset [235] Japanese Biography scans (Hw, Pr) IR ARDIS [236] Swedish Digit (Hw) DR 18-19th documents like business letters or government PDFs, or medical documents like Diagnostic reports, pathology papers etc.…”

On Optimality of Long Document Classification using Deep Learning

“…Several word embeddings are believed to be true in classifying the document [4]. As words cannot be relevant in all contexts, except they are perceived with dictionary meaning and related transitives, a word in a sentence will have the exact meaning in the context.…”