New Technology Helps Visually Impaired Grocery Shop

At the University of California, San Diego (UCSD), researchers bring two great ideas together to form a third that will aid individuals who are visually impaired to do their grocery shopping unaided.

The GroZi and CAPTCHA are being united through the newly conceived Soylent grid by a team of computer scientists led by professor Serge Belongie of UCSD. The GroZi project inspired the design for the Soylent grid and will operate via CAPTCHA technology.

The GroZi is a wearable computer with an incorporated camera that uses assistive technology to create a personal shopping aid for the visually impaired. The GroZi can lead a visually impaired person to a desired product by performing a computer analysis of the live video stream.

CAPTCHA technology is the spam filtering technology that Internet users come across when a computer program system wants to confirm that a user is a live and real person: you are requested to replicate a random series of distorted numbers and letters (with a second chance for when they are distorted too greatly). Web site hosting companies and other Internet businesses need to take advantage of CAPTCHA spam filtering capabilities to protect their interests and their clients. So at any moment in the Internet world a potential thousands of people are typing in nonsensical random characters for spam security clearance.

The Soylent grid is a new technological concept that links the GroZi to CAPTCHA spam filtering technology. The Soylent receives the video stream from the GroZi and computer vision algorithms solve the video stream object identification problem. Or in a second scenario, an Internet user encountering a CAPTCHA filter that is in the Soylent grid, which is receiving the live video stream(s), receive the lie stream through the Soylent grid and immediately solve the object identification problem and instantaneously send the object identity information back through the Soylent grid to the GroZi to guide the shopper. In this way Internet users "can help label still shots from live video streams in real time," as Belongie said.