Sunnyvale, CA (PRWEB) April 07, 2014
Today Multicoreware is announcing the availability of custom professional video applications leveraging Deep Learning Neural Networks and advanced machine learning techniques. These new technologies are enabling human quality and superhuman competitive classification for many aspects of video QC. Combining these systems with CUDA and OpenCL enabled GPU's makes real time detection of banding, macro-blocking, and digital glitches a reality. Accuracy rates in the high 90's have been achieved for feature detection.
Dr. Wen-Mei Hwu, CTO of MulticoreWare explained,"MulticoreWare has focused its world-class R&D assets in Urbana-Champaign to develop a leading implementation of Neural Networks and Deep Learning for real-time video analytics. In doing so, we are able to provide incredible speed, scale and accuracy in real-time. Specifically, for the professional video industry, we are enabling real-time, highly accurate identification and recognition of many features and objects. We are proud to introduce this technology at the National Association of Broadcasters show, 2014."
“We have worked intensively to develop leading Neural Network techniques to deliver the highest human and machine visible video quality for professional video applications. To achieve leading performance, we optimized code for heterogeneous multicore environments, with full support for CUDA and OpenCL standards on many platforms. We invite industry leaders to see our demo at NAB 2014,” said Curtis Davis, co-founder, VP of engineering and COO of MulticoreWare.
MulticoreWare develops high performance video processing libraries including the x265 HEVC video encoder and the UHDcode HEVC video decoder. MulticoreWare has the largest Open CL development team in the industry, developing accelerated software solutions for mobile devices, PCs and embedded devices.
For more information, contact sales(at)multicorewareinc(dot)com, or visit MulticoreWare on the web at http://www.multicorewareinc.com.