Researchers have determined that individuals can be identified by the footage from their personal mounted cameras, according to an article by James Vincent for The Verge. The Hebrew University of Jerusalem researchers, Yedid Hoshen and Professor Shmuel Peleg, say that person’s movement is characteristic and unique enough that its effects on the footage from a head- or body-mounted camera is distinguishable and can be used for identification purposes.
In a paper on their study, entitled Egocentric Video Biometrics, the researchers note that they could identify volunteers from only four seconds of video footage, by tracking objects’ “optical flow” through the frame of the video. As noted in an Ars Technica article by Megan Geuss, the researchers made use of Convolutional Neural Networks (CNN), a means of training an algorithm that entails a process “of deep learning that looks at layers of an image starting with small pieces and moving up to larger sections of the image.”
The concept of using involuntary body movements for authentication purposes isn’t as far-fetched as it might seem to some; the new Arki wristband is a fitness-tracking device designed to monitor the user’s gait and posture, and its makers say it can also be used for identification purposes. This is essentially the same technological process used by Hoshen and Prof. Peleg – gait biometrics.
Other organizations have been looking into the metaphorical fingerprints unconsciously left by users just going about their routine activity, as in the case of the Raytheon engineer’s concept for a mouse that only works via biometric authentication of a user’s unique grip; and, more broadly, there is a startup looking into identity authentication via users’ online activity on social networks and other digital applications.
As for the Hebrew University of Jerusalem researchers, they are modest in the ambitions of their study; Prof. Peleg says that their aim was merely to “make people think twice before uploading egocentric footage to the web.” Still, the technique could have major applications, given the recent interest in outfitting police with body cameras, and the growing popularity of wearable tech devices like Google Glass.