The Artistic-Faces dataset contains 160 artistic portraits of 16 different artists, which were chosen to be representative of a wide range of artistic styles, both in geometry and texture. It contains artwork styles ranging from High Renaissance through Cubism to Comics.
The portraits are annotated with 68 facial landmarks to remain consistent with previous works in facial landmark detection of natural faces. For each artwork we provide the following metadata : artist name, artwork title, style, date and source.
Our dataset is mainly drawn from the Painter By Numbers (PBN) dataset, which consists of 103,250 artworks. The images in the PBN dataset were mostly obtained from the WikiArt. In addition, our dataset contains images from online art collections provided by: Van Gogh Museum's collection, Image-Duplicator by the Roy Lichtenstein Foundation, National Gallery of Art (NGA) image collection, Tate collection, The Met collection, Nasjonalmuseet collection and Google image search engine.
Notice that some of the images in the Artistic Faces Dataset are protected by copyrights. Please utilize these images only for the purposes of data mining, which constitutes as a form of fair use. You can find copyright details in the metadata file.