Project Contributors: Dirk Kayitare, Ngan Nguyen, Pratham Singhal, Sophie Smith, Katherine Zelaya

Community Partner: 1. Haiti Institute and 2. Ancient Art Archive

Collaborating with the Haiti Institute and the Ancient Art Archive, the Sewanee DataLab photos team applied machine learning techniques to curate photographic evidence of the impact of Sewanee’s work in Haiti.

Problem: 

  1. No conclusions or insights have been drawn from the files of photos from Haiti.
  2. Rock art is continuously threatened by people and the environment, and as the rock art disappears so do the stories behind them.

Solution: 

  1. Convolutional neural networks were used to quantify objects and humans in the photos, drawing necessary conclusions and insights from the images.
  2. Convolutional neural networks were used to label and digitize rock art objects, allowing the rock art and its stories to be safely stored.

Impact:

  1. The work from the summer measured economic development and informed conversations about development in the community.
  2. The work also created statistical tools for research and preservation of heritage sites.