вторник, июля 09, 2019
Mapping a Book by its Cover
The Pudding has created an interactive map showing the covers of 5,000 books. All 5,000 books mapped on 11 Years of Top-Selling Book Covers, Arranged by Visual Similarity appeared on the New York Times' 'Best Selling' or 'Also Selling' lists since June 2008. The book covers are organized on the map based on their visual similarity, as judged and organized by machine learning.
Color seems to play a very prominent role in determining 'visual similarity' in the machine algorithm used by The Pudding. If you zoom out so that you can see all 5,000 book covers you can see that a lot of the grouping and organization appears to be strongly influenced by the dominant color of each book.
The Pudding's map comes with a number of filters which allow you to explore the book covers by genre and by visual motif. The visual motif filter allows you to highlight on the map images which contain 'faces', 'landscapes', 'smiles' etc. Therefore the motif filter provides another way to explore the book covers by visual similarity. The genre filter is a great way to explore and examine the visual trends used by publishers for different genres of literature.
The map itself was created using OpenSeadragon, a library for viewing high resolution zoomable images. OpenSeadragon supports a number of different image tile services, including WMS and IIIF.
Подписаться на:
Комментарии к сообщению (Atom)
Комментариев нет:
Отправить комментарий