Add dataset: nls_chapbooks_illustrations

[davanstrien]

A URL for this dataset

https://gitlab.com/vgg/nls-chapbooks-illustrations

Dataset description

This proposed dataset consists of a few components:

• Chapbooks printed in Scotland: a dataset consisting of "more than 3,000 chapbooks printed in Scotland. They form part of the Lauriston Castle Collection, which was bequeathed to the Library in 1926. It includes some 500 chapbook volumes containing around 5,500 individual items, more than half of which were printed in Scotland."
• annotations of visual content in these Chapbooks. These annotations consist of COCO bounding boxes for the visual content contained in these Chapbooks.

This is an excellent dataset for training or evaluating object detection models on historical material. Identifying visual content in digitised material is very useful for both LAM institutions and researchers. In addition, the dataset is also relatively large compared to many LAM object detection training datasets, which are sometimes only large enough to perform evaluation and not large enough for training (particularly without transfer learning).

Dataset modality

Image

Dataset licence

Creative Commons Public Domain Dedication and Certification

Other licence

No response

How can you access this data

As a download from a repository/website

Confirm the dataset has an open licence

• [X] To the best of my knowledge, this dataset is accessible via an open licence

Contact details for data custodian

No response

[davanstrien]

Initial data loading script here: https://huggingface.co/datasets/biglam/nls_chapbook_illustrations. Needs some tidying still. In contact with authors/NLS about working on a datacard.

[davanstrien]

Closing this one for now