A new project being developed for Google Glass aims to identify people based on the clothing and accessories they wear. Once a familiar person has been spotted in the crowd, Glass will then overlay their name above them on the display like you’d see in a video game.
The project, partially funded by Google, was recently showcased at the HotMobile technology conference in Jekyll Island, Georgia. It is designed to help people find friends or family in large, crowded areas like airports, sports stadiums and shopping malls.
InSight developer Srihari Nelakuditi at the University of South Carolina in Columbia said they couldn’t use facial recognition as it would be unlikely that the person you are trying to find is looking directly at the visor’s camera. Instead, they developed a program that uses a “fashion fingerprint” of the clothes and accessories someone is wearing to help spot them later on.
This digital fashion fingerprint is built using a smartphone app that snaps a series of pictures of a particular individual. These photos are then used to create a file called a spatiogram that captures spatial distribution of colors, patterns and textures of the clothing the subject has on. Early testing using 15 volunteers found the system was able to locate the target 93 percent of the time.
Because the system profiles clothing and accessories instead of human faces, it’s said to protect peoples' privacy as the fingerprint changes every time a person changes their clothes.