Letter: Video analysis research at Duke

I would like to make a clarification and an apology concerning a recent Chronicle article on recording thousands of students' faces.


I have never worked on facial recognition. We recorded the data to research methods to analyze the motion of objects in video, whether they are people, cars, fish or other. My students and I have also done research on diagnosing retinopathy of prematurity in infants, lower the cost of screening for colorectal cancer, determine the provenance of documents on Golden Age Spanish theater from handwriting analysis, track hand motion to interpret American Sign Language and much more. We were and are uninterested in determining personal identity from images. The closest we came to a project that involves personal identity was a method for protecting sensitive personal information from the prying eyes of surveillance cameras. I will explain in a separate guest column what we do do in our work on motion analysis in video.


I genuinely thought I was following IRB [Institutional Review Board] guidelines in our project on video analysis. As directed, we placed posters at all entrance points to the recording area. The posters described our research and its goals, and gave people contact information (email, phone) they could use to request for any data to be erased. Nobody did.

However, it turns out that I deviated from IRB guidance in two aspects: Recording outdoors rather than indoors, and making the data available without protections. IRB is not to be blamed, as I failed to consult them at critical junctures.

I take full responsibility for my mistakes, and I apologize to all people who were recorded and to Duke for their consequences.

Editor's Note: This letter has been updated to include information about an upcoming guest column.


Share and discuss “Letter: Video analysis research at Duke” on social media.