Start-up Aims for AI-based Real-time Video Identification

Valossa, a start-up based in Oulu, Finland, is working with Orange Silicon Valley and EchoStreams to demonstrate its AI-based real-time video identification technology.

Valossa’s Val.ai is based on years of research at the world-renowned computer vision and AI labs of the University of Oulu. The company says its technology analyzes streaming videos in real-time to identify thousands of concepts (e.g., places, objects and unique topics).

A system powered by EchoStreams FlacheSAN1N NVMe storage technology and GS420ND4 Exascale platform runs Valossa's software for identifying objects and people in real-time across more than 25 simultaneous HD video streams. The optimized-for-video supercomputer uses 16 state-of-the-art NVIDIA GPUs, exceeding 48,000 CUDA cores in a single 4U server.

“We are excited to collaborate with Valossa and OSV to bring High Performance Server and Storage platforms, which normally are only available in Supercomputing Data Centers, to the Media and Entertainment industry,” said Andy Lee, EchoStreams Director of Product Marketing. “By harnessing HPC computing power and Valossa AI, this demonstration is truly a breakthrough implementation of the Artificial Intelligence concept in real-world applications such as the Media and Broadcasting industry.”

http://www.valossa.com