Topic: The potential of computer vision tools for improving demographic measurement in low-income countries
Date: Monday, December 10, 2018
Time: 12:00 pm - 1:00 pm
Location: 1101 Morrill Hall
Speaker: Stephane Helleringer, Johns Hopkins
Host: Maryland Population Research Center (MPRC)
Dr. Helleringer will present the results of a recent trial of a a computer vision approach to improving age data in low-income countries. In such settings, age is often misreported during surveys and censuses due to inadequate knowledge of dates of birth, low numeracy levels in some population sub-groups, or deliberate behaviors of data collectors. This may lead to (sometimes severe) bias in estimates of key demographic indicators such as life expectancy or the total fertility rate. To address this issue, Dr. Helleringer and colleagues trained a computer system to automatically produce an age estimate from a single photograph of the face of study participants. That system was then evaluated in a demographic surveillance system in Senegal (West Africa), where gold standard age data are available. Other potential applications of computer vision in demographic and health studies will be discussed.
