SHIBR- The Swedish Historical Birth Records: A Semi-Annotated Dataset (SHIBR)
Historical data of birth records, Handwritten documents, Public dataset, Word spotting
This dataset is taken from the Arkiv Digital AD AB image and index database. When a child was born he or she was registered in a church record book called Birth and Christening records by the priest. They registered the name of the child, when the child was born and baptized, where the child was living and information about the father and mother of the child. The index is based on manual annotation of images from several books between the year 1800 to 1840.
The dataset consists of 191,301 index rows and 15,000 images and has been divided into
train: 133,941 index rows and 10,500 images
eval: 28,303 index rows and 2,250 images
test: 29,057 index rows and 2,250 images