The dataset I’ve chosen isĀ Japanese woodblock prints (ukiyo-e). The images from this dataset consist of landscapes, actors, close-ups of people, and scenes from history. I am most interested by the landscapes and the close-ups with interesting facial expressions. Some contributors to the dataset have metadata including title, date, artist and some elements found in the images.


For my project I would like to either a) caption theĀ images with a saying, based on the elements of the image (generate text) or b) analyze the expressions on people’s faces and categorize them.