Embedded Vision And Android 4: More Features GaloreOctober 19, 2011
After delaying the introduction of "Ice Cream Sandwich" one week out of respect for Steve Jobs' passing, as well as relocating the event from San Diego to Hong Kong, Google unveiled Android 4.0 last night in conjunction with showcasing its premier hardware implementation, the Samsung-designed Galaxy Nexus. Ironically, considering that just a week ago I wrote about facial recognition-based handset unlock schemes on Nokia's legacy Symbian O/S, "Ice Cream Sandwich" touts a near-identical Face Unlock implementation (although it reportedly didn't work too well when Google's Matias Duarte attempted to demonstrate it). Engadget subsequently gave the feature a whirl, with a more positive outcome:
Later that same evening, Google's Andy Rubin sat down with Walt Mossberg from the Wall Street Journal and revealed additional Face Unlock details. The core technology come from Google's earlier acquisition of Pittsburgh Pattern Recognition ("PittPatt"), which I wrote about in late July. A notable percentage of the facial recognition algorithm runs on the GPU, not the CPU, as a GPGPU (general-purpose computing on graphics processing units) implementation showcase. And speaking of faces, Google has also beefed up Android 4.0's built-in photography utility, among other things including the ability to detect and adjust focus and exposure based on the presence of up to 10 visages in the frame.