Image: 4,900 images from SeMA's digital collection. Features extracted using the Keras implementation of Inception-v3, dimension-reduced using PCA and t-SNE, visualized into grid using Rasterfairy.

Image: 4,900 images from SeMA’s digital collection. Features extracted using the Keras implementation of Inception-v3, dimension-reduced using PCA and t-SNE, visualized into grid using Rasterfairy.

Image: 4,900 images from SeMA's digital collection; features extracted using Nasnet, visualized using PCA and t-SNE (library: Bokeh).

Same dataset; features extracted using Nasnet, visualized using PCA and t-SNE (library: Bokeh). Click image for interactive version.