Now, as investigators pore through the thousands and thousands of images, searching for anything even remotely suspicious, they are enlisting the aid of software like that developed by Dartmouth’s Lorenzo Torresani, assistant professor of computer science. Such software allows a user to plug in attributes such as “brown hair” and search the files for anyone of that description, says NBC News.
“You can enter these attributes and the (software) builds a visual model of the target and then scans for the pattern,” Torresani tells NBC News.
NBC News also spoke with Dartmouth Professor of Computer Science Hany Farid about the possibility that some of these images could be fakes.
Farid, co-founder of Fourandsix, a startup that developed image-authentication software that reads the JPEG “fingerprint” of an image, says the software can tell investigators that something has been altered, even if it cannot say exactly what was altered. “You can think of it as a triage,” he says.
Read the full story, published 4/16/13 by NBC News.
