The Yale Digital Humanities Lab presents Neural Neighbors, a project intended to enable users to "identify visual similarity at scale across collections of photographs and videos." Currently, the project uses a subset selected from the Beinecke Library's Meserve-Kunhardt Collection, which was assembled by Frederick Hill Meserve, who collected photographs between 1897 and his death in 1962. This large collection documents United States history, especially the American Civil War, and also includes portraits of some 6,500 individuals such as Abraham Lincoln, Ulysses S. Grant, U.S. and Confederate military and governmental personnel, cabinet members, and authors, poets, clergy, actors, actresses, and other entertainers. When Neural Neighbors loads, viewers first see a screen full of images. Mousing over any image brings up additional, visually related images. For example, an untitled, familiar-looking portrait of Lincoln in a seated position is associated with similarly posed portraits of (among others) George Armstrong Halsey (1827-1894) and P. T. Barnum (1810-1891). The experimental nature of the project is revealed at this point, however, because it's not possible to retrieve more information about the images without leaving Neural Neighbors and searching Yale's collections, which can be a frustrating exercise.
(no comments available yet for this resource)