Google launched a few interesting products as well as feature upgrades in the last few weeks. First of all, Google Gears was launced officially at the Google DevDay 2007. Google Gears is an open source technology for creating offline web applications and it accomplishes this by installing a local database store that can communicate with the applications running on our local browsers. Second of all, it looks like Google may have silently slipped in facial recognition features into its image search, as a logical extension to the recent upgrades it made to Picasa after purchasing Neven Vision.

Google Gears uses Javascript APIs to store and serve application resources locally through a fully-searchable relational database. The API includes, beyond other things, support for asynchronous Javascript calls which might make the synchronization and the toggle between offline and online states a lot less painful to the end user of the application. Right now, Google Gears is being supported for Firefox 1.5+ and Internet Explorer 6.0+ under the Windows XP/Vista, Mac OS X 10.2+ as well as the Linux machines (excluding Fedora Core 6) with glibc 2.3.5 or higher and libstdc++5. A look at the developer forum for Google Gears suggests some pending problems with installation and support for multiple platforms but Google gears is definitely a giant step in the direction of transforming the web into the next generation equivalent of an Operating System and wins my vote :).

As far as the facial recognition idea goes, notice how an image search for my name shows a bunch of results that are not always faces. But, add a little “&imgtype=face” to the end of the URL and voila! the new results contain only the faces ! Sadly, none of the two faces is me even though they had an undoubted impact on my technical break-a-back nights at Grad school! I just hope this face recognition feature will one day smarten up to recognize the right face for the right image search and that it could eventually be expanded to other contexts as well …:)

Hopefully, they could move some of the Google search statistics to the local machine using Google gears and help address some of those privacy concerns but that is a long shot in the dark.

p.s: One of the people in the picture is my professor Dr. Mansoor Alam while the other is my friend Sachin Shetty, with whom, I worked on and wrote the paper on evolutionary scheduling in gigabit networks.