A Century of Portraits

decade average images for the 20th century — Average images of high school students for each decade of the 20th century.

Abstract

Many details about our world are not captured in written records because they are too mundane or too abstract to describe in words. Fortunately, since the invention of the camera, an ever-increasing number of photographs capture much of this otherwise lost information. This plethora of artifacts documenting our “visual culture” is a treasure trove of knowledge as yet untapped by historians. We present a dataset of 37,921 frontal-facing American high school yearbook photos that allow us to use computation to glimpse into the historical visual record too voluminous to be evaluated manually. The collected portraits provide a constant visual frame of reference with varying content. We can therefore use them to consider issues such as a decade’s defining style elements, or trends in fashion and social norms over time. We demonstrate that our historical image dataset may be used together with weakly-supervised data-driven techniques to perform scalable historical analysis of large image corpora with minimal human effort, much in the same way that large text corpora together with natural lan- guage processing revolutionized historians’ workflow. Furthermore, we demonstrate the use of our dataset in dating grayscale portraits using deep learning methods.

People

Paper

A Century of Portraits: A Visual Historical Record of American High School Yearbooks

Shiry Ginosar, Kate Rakelly, Sarah Sachs, Brian Yin, Crystal Lee, Philipp Krähenbühl, Alexei A. Efros. A Century of Portraits: A Visual Historical Record of American High School Yearbooks, Extreme Imaging Workshop, International Conference on Computer Vision, ICCV 2015. and IEEE Transactions on Computational Imaging, September 2017. PDF, BibTeX

@ARTICLE{ginosar2017yearbooks,
  author={S. Ginosar and K. Rakelly and S. M. Sachs and B. Yin and C. Lee and P. Krähenbühl and A. A. Efros},
  journal={IEEE Transactions on Computational Imaging},
  title={A Century of Portraits: A Visual Historical Record of American High School Yearbooks},
  year={2017},
  volume={3},
  number={3},
  pages={421-431},
  keywords={Data mining;Face;Imaging;Market research;Sociology;Statistics;Visualization;Data mining;deep learning;historical data;image dating},
  doi={10.1109/TCI.2017.2699865},
  month={Sept}
}

Sample Results

The Top Trends of Each Decade

top female hairstyle trends for the 1960s and 1970s — Discriminative clusters of high school girls’ styles from the 1960's and 1970's. The left-most entry in each row displays the cluster average. Note that the clusters correspond to the quintessential hair and accessory styles of each decade: The bob, “winged” flip, bubble cut and signature glasses of the 60's. The long hair, Afros and bouffants of the 70's. These fashions emerge from the data in a weakly-supervised, data-driven process.

smiles over the years — Smiles increasing over time, but women always smile more than men: Male and female Average lip curvature by year with one STD error bars.

Dataset

The Yearbook Dataset

The Yearbook Dataset of frontal-facing American high-school seniors from 1905 to 2013 is hosted on space donated by Dropbox. All faces are aligned using an affine transformation in a process described in the paper.

The training and test lists for female faces used in the paper are also provided.

If you would like to obtain other formats of the data (raw images, non-frontal facing portraits etc) or, alternatively, if you would like to contribute more yearbook data send us an email to: <shiry at eecs dot berkeley dot edu>.

This dataset is covered by the MIT license.