Project Dataface began as a joint research project for an ACES seminar between Christian Johnson and Jeremy Krach, both Class of 2017. The original project attempted to raise awareness about digital privacy, specifically identifying individuals through images on social media.
Johnson and Krach developed a tool to harvest profile pictures on Facebook from more than 20,000 University of Maryland students. Using the profile pictures database, they planned to create a Google Glass application to recognize anyone on campus. Once someone is recognized, the application would create a post on social media, such as a Tweet or a Facebook post. By having a stranger post directly to their profile noting their location, the application would generate some discomfort and hopefully encourage discussions on publicly available information.
Johnson and Krach explored different facial recognition algorithms, but found that all potential algorithms required more than seven images to reach at least a 90 percent likelihood of correctly finding a match and Facebook only provided one image per person for their database.
For their seminar project Johnson and Krach proposed that instead of using the Internet to collect images, they would collect images in real life and then search the Internet to find matches to identify individuals.
Krach later continued this research individually, using Google Glass to experiment with different picture collecting methods that would maximize battery life and minimize data loss.
Krach has found he can reliably collect more than 90 percent of the images taken by Glass and has maximized the battery life to more than 50 minutes.
The battery life of the Google Glass is tied to the hardware and is not affected by the software collecting images. As better hardware is developed, Krach’s process will scale to enable massive data collection schemes.
As he completes the backend, Krach wants to “democratize” surveillance and allow any person with a Google Glass to create a recognition or surveillance system, changing the discourse on public surveillance.