New software helps prevent leakage of sensitive info on camera

Scientists, including one of Indian-origin, have developed a software that helps prevent accidental disclosure of trade secrets and other restricted information within a camera's field of view, by letting users specify what others can see.
A video chat with collaborators, customers or suppliers outside of the office, for example, could show confidential product plans drawn on a white-board in the background, or sales figures or source code on nearby computer screens.
Using a smartphone to scan a receipt for expense purposes could also expose portions of meeting notes, pill bottles, and other personal items on the user's desk, researchers said.

"There are more and more cameras every year. They are incredibly useful," said Landon Cox from Duke University in the US.
"But the downside is we are now converting large swaths of our surroundings to a digital format that is easy to access and share, including things we might not want to be digitising," said Cox.
The simplest way to ensure privacy is to disable the camera or microphone when sensitive information is in the frame, he said.

But rather than all or nothing, researchers wanted to give users more granular control over which objects in the camera's view are shared and which are kept private.

Prior efforts to safeguard confidential information in photos and video use a "blacklist" approach. Developers anticipate things that users might want to hide, and build software that blurs or masks them in each frame.
But coming up with an exhaustive list of potentially troublesome objects is virtually impossible, researchers said.
"Things that some people consider sensitive might not be sensitive to you. It is hard to build something that covers all possible scenarios," said Ashwin Machanavajjhala from Duke University.
Instead of relying on a developer's best "guesstimate" of which objects should be "public" and which should be "private," researchers set things up so that the user makes that determination. And instead of choosing what to hide, the user chooses what to show.

One example of the approach is designed to protect sensitive information on two-dimensional surfaces such as white-boards and computer presentation slides.
The other approach safeguards images of three-dimensional objects such as keyboards and faces.
In both cases, users select the part of a scene that is 'OK' to share by drawing a rectangular border around it, either by hand or with a few clicks of a mouse.
Once it knows what it is looking for, the software intercepts all incoming frames from the video stream and rapidly scans frame by frame for a match using computer vision technology, researchers said.

Only authorised objects are allowed to pass from the camera to third-party software, like smartphone apps. Everything else is blocked out by default, they said.