Tag Archives: Google

And another story on 60 Minutes about a major personal mystery which was solved by visual recognition and visual memory

Memories of the scenery of a long-ago journey, Google Earth, a very old photograph of a young boy and face recognition – these are the elements that found the solution to an impossible quest. Saroo Brierley’s amazing story of finding a lost mother within the incomprehensibly huge population and landmass of India is a demonstration of the incredible power and potential of the use of images as data in contemporary computer technology and the natural visual processing and visual memory capabilities of the human brain.

Lost and Found. 60 Minutes (Australia). June 21st 2013. http://sixtyminutes.ninemsn.com.au/article.aspx?id=8678261

Saroo was also a guest on Breakfast on ABC’s Radio National on June 25th 2013, interviewed with her usual skill and intelligence by Fran Kelly. Saroo is promoting his autobiography titled A Long Way Home.

Long Way Home: An extraordinary story. Breakfast. http://www.abc.net.au/radionational/programs/breakfast/long-way-home-an-extraordinary-story/4777562

Fashion recognition app that could be useful for prosopagnosics

Google Glass app identifies you by your fashion sense. by Paul Marks New Scientist. Magazine issue 2907, 7 March 2013.

I’m not completely clear how this recognition technology works, but it says it creates a spatiogram that is a record of what a person is wearing, including colours, and it can be used to identify an individual in a crowd. I found it interesting that face recognition was dismissed as unfeasable, considering the many articles I’ve read over the years making big claims for face recognition technology:

“Face recognition systems cannot be used for this, says InSight developer Srihari Nelakuditi at the University of South Carolina in Columbia, because it is unlikely someone in a crowd will be looking straight at a headset’s camera.”

Unsolved Mysteries is my guilty TV watching pleasure, but I read New Scientist with pride

Journalism in the areas of crime, the supernatural and miscellaneous weird stuff are not my usual choices in reading or viewing, at least not in the daytime, but there’s nothing more fascinating than a mystery, except for a clever solution to a mystery. One interesting aspect of this compelling TV show from the United States, which is generally broadcast late at night around the weekend, is that every episode of Unsolved Mysteries involves facial recognition as the solution or an important element of the story’s mystery. Other types of visual recognition can be an important feature in the narratives. One episode of the show recently broadcast in Australia was a murder mystery in which a police officer who had just investigated a murder later attended the home of the victim’s girlfriend who had disappeared. Just by chance the police officer looked into a linen closet and noticed in there pillow-slips with a fabric design which matched the sheet that had been found wrapped around the boyfriend’s body. I’ll bet that’s a variety of visual recognition that the scientists haven’t named yet.

While catching up with reading some back issues of New Scientist magazine today I came across another story about a criminal conviction that resulted from some very sharp soft-furnishing fabric design recognition skills on the part of an American law-enforcement officer. It’s not a nice story, not nice at all, but at least there’s some inspiration to be found in the good people using technology to fight the vile crime of child sexual abuse. An investigator at the National Center for Missing and Exploited Children noticed that among the countless horrible images received at the NCMEC two were of girls of a similar age on what looked like the same bedspread of a distinctive appearance. I have no idea how the police trace these things, but the locations where that style of bedspread were sold were identified, and this was the clue that led to the identification of the children and the criminal. Google have developed for the NCMEC software designed to achieve similar feats of visual object recognition as the investigator’s human visual recognition of the bedspread. It is hoped that the automation of the identification of items of interior decoration in images of child abuse will help to solve more crimes. Of course, the NCMEC also works to identify the child victims of crime themselves, in the Child Victim Identification Program. Google, Yahoo and Microsoft are partners in the NCMEC’s Technology Coalition, and the application of technology to the task of identification is viewed as the only way to deal with the increasing volume of pornographic material submitted to the NCMEC every year.

Unsolved Mysteries   http://www.unsolved.com/

Peter Aldhous Fighting online child porn. New Scientist. April 9th 2011. p.23-24. http://www.newscientist.com/article/mg21028075.000-automating-the-hunt-for-child-pornographers.html

National Center for Missing & Exploited Children (US)  http://www.missingkids.com/missingkids/servlet/PublicHomeServlet?