NVIDIA CUDA Technology To Dramatically Speed Up Facial Recognition in CyberLink MediaShow 5

You all recall Intel's interesting "Deep Viewer" tech demo that we showed you way back in January, live on the CES show floor.  Of course you do.  Well it appears that NVIDIA has cooked up a solution with the folks at CyberLink to offer at least some competitive technology to what Intel has thus far only used as a showcase for their multithreaded processor technology.  CyberLink's MediaShow 5 now features facial recognition technology in order to help find those creamed-squash smeared baby pictures that much faster.  Here's a taste...


Smart Face Tagging with CyberLink MediaShow 5

SANTA CLARA, Calif.—Sept. 8, 2009— NVIDIA Corp. announced today that CyberLink MediaShow 5, a new software program that organizes digital photos based on who is in them, is utilizing the CUDA parallel processing power of NVIDIA® GeForce® graphics processing units (GPUs) to  search and sort photo libraries. 
 
CyberLink MediaShow 5 searches and organizes your photo collection using a sophisticated face recognition algorithm.  By utilizing co-processing – which leverages the massive parallel processing power of NVIDIA GeForce GPUs in tandem with the computer’s central processor (CPU) – CyberLink MediaShow 5 provides up to a 70% improvement in application speed so you can find the photos you want faster.
 
“Consumers are accumulating massive photo libraries, given the proliferation of digital cameras, camcorders and smart phones,” said Alice H. Chang, CEO of CyberLink. “MediaShow 5 helps them better manage their photo collections by searching photos based on face recognition and tagging. NVIDIA CUDA technology significantly reduces the processing time involved in this.”
 
In addition to face recognition, MediaShow 5 uses NVIDIA® CUDA™ technology to accelerate video conversion while still delivering ultra-high image quality. 


Hit the full release at the via link on the bottom here if you like but in short, the software will still function on your machine with the tagging functionality enabled.  However, the claim here is that you may just poke along a bit slower if there isn't a GeForce GPU under your hood.