Many details about our world are not captured in written records because theyare too mundane or too abstract to describe in words. Fortunately, since theinvention of the camera, an ever-increasing number of photographs capture muchof this otherwise lost information. This plethora of artifacts documenting our"visual culture" is a treasure trove of knowledge as yet untapped byhistorians. We present a dataset of 37,921 frontal-facing American high schoolyearbook photos that allow us to use computation to glimpse into the historicalvisual record too voluminous to be evaluated manually. The collected portraitsprovide a constant visual frame of reference with varying content. We cantherefore use them to consider issues such as a decade's defining styleelements, or trends in fashion and social norms over time. We demonstrate thatour historical image dataset may be used together with weakly-superviseddata-driven techniques to perform scalable historical analysis of large imagecorpora with minimal human effort, much in the same way that large text corporatogether with natural language processing revolutionized historians' workflow.Furthermore, we demonstrate the use of our dataset in dating grayscaleportraits using deep learning methods.
展开▼
