Extreme Imaging


July 18, 2017

robert pless

Cameras have become ubiquitous. Security cameras increasingly dot our landscapes, and smart phones mean that most of us have a camera at the ready at all times.

Dr. Robert Pless foresees the myriad beneficial applications of the images these cameras capture, and he is capitalizing on them to help address national needs and social justice issues.

“The social goal is to use the fact that there are so many cameras around the world in so many people’s hands and they see so many important things,” Dr. Pless argues. “By connecting them to each other or central databases we can find new ways of attacking really important problems.”

One of the problems that he and his research group have been working on is an app called Traffickcam, which is now being used to help law enforcement track the locations of hotels used in sex-trafficking of minors. Sex-traffickers often take photos of these children in hotels and post them online. Having detailed photos of hotel rooms across the country that law enforcement can match against the traffickers’ photos can help law enforcement find the hotels where victims may have been taken.

Traffickcam allows travelers to take photos of their hotel rooms and upload them to the database that law enforcement uses. Since it was launched six months ago, travelers have contributed more than two million photos.

Another longer-term project that Dr. Pless works on takes in vast numbers of images—far more than the Traffickcam project—to record how the world has been changing. For the past 10 years, AMOS (Archive of Many Outdoor Scenes) has logged images from every publicly available webcam that his group could find. It records one photo every half hour from each of the approximately 35,000 cameras posted around the world on campuses, beaches, streets, and more. It currently has a library of one billion images.

What researchers can learn from this library is almost limitless. Dr. Pless currently is using the images to discover, for example, how many more people are using bike lanes in Washington, D.C., and how vegetation is changing over time in other areas. To do this successfully, he develops algorithms that can manipulate the vast quantities of data stored in extremely large photo collections, so new information can be learned from them.

“I’m exploring the limits of computational photography, which is when you have a computer attached to the camera taking pictures,” says Dr. Pless. “What’s unique about my research compared to my colleagues working in computational photography is that I think about extreme imaging questions, whether it’s time lapses over huge amounts of time or pictures from huge numbers of cameras.”