KWS

Handwritting Keyword Indexing and Search
PRHLT Demonstrators

EXPERIMENTAL, SMALL/MEDIUM SIZE DATASETS, most with ground truth (GT):

Various collections indexed together (HMM optical models, lexicon-based)

Plantas - 739 pages Spanish botanical book (HMM optical models, lexicon-based, no GT)

Bentham 83 pages (HMM optical models, lexicon-based, with GT -- see LARGE SCALE below for the full collection)

Jane Austen 128 pages (HMM optical models, lexicon-based, with GT)

Wiensanktulrich 52 pages, in German (HMM optical models, lexicon-based, with GT)

La contienda 178 pages (HMM optical models, lexicon-based, with GT -- a small part of TSO, see LARGE SCALE below)

Passau miscellanea (CRNN optical models, lexicon-free, with GT)

Chancery-Guerin 95 Acts (CRNN optical models, lexicon-free, with GT -- see LARGE SCALE below for the full collection)

Chancery-Guerin 444 Acts (CRNN optical models, lexicon-free, no GT -- see LARGE SCALE below for the full collection)

Chancery-Guerin lemmatized 444 Acts (CRNN optical models, lexicon-free, no GT -- see LARGE SCALE below)

Carabela - 340 images of interest to underwater archaeology (CRNN opt. mod., lex-free, GT -- see LARGE SCALE below)

NEW: Hyphenated-word indexing, FCR 200 images, 310 pages (CRNN, opt. mod., lex-free, GT -- see LARGE SCALE below)

NEW - SQL-like Tabular Queries in Hndwritten Table Images:

Small set of Bistums Passau regisiters, including 43 varied table images (CRNN optical models)

Handwritten Music Symbol Indexing and Symbol-Sequence Search:

Vorau-253 Music manuscript 490 page images (CRNN optical models, 44 images with partial GT)

LARGE SCALE DATASETS (all CRNN optical models, lexicon-free, no GT):

Chancery - Trésor des Chartes registers (199 manuscripts, 82,000 page images)

FCR - Finnish Court Records (125 units, 102,000 images, about 140,000 handwritten pages)

Bentham Papers (193 boxes, 90,000 handwritten page images) -- NEW, RICH-FEATURED SEARCH ENGINE (help)

Teatro del Siglo de Oro - Spanish Golden Eage Theatre (328 manuscripts, 41,000 page images)

Carabela - manuscripts valuable to underwater archaeology (328 folders, 31,000 imgs., abstruse scripts -- work in progress)