Hybrid codec module extracts video data for cloud AI

May 17, 2018 // By Julien Happich
In addition to decoding and encoding multi-channel, high-resolution, high-quality image and video data, Socionext's M820L hybrid codec built around dedicated codec hardware and flexible software processing supports advanced video data processing capabilities such as the extraction of information from a video to integrate it into the output data.

Assembled into a PCIe low profile card, the M820L can be inserted into the PCIe slot of a PC or a server and functions as an accelerator to reduce the load of a CPU's encoding processing. Socionext has collaborated with Fujitsu Limited to develop a video analysis system in combination with the "Fujitsu Human-Centric AI Zinrai". The M820L receives signals from cameras and converts them to highly compressed, high-quality image data. In parallel, it extracts information for video processing and analysis, and inputs such data to the Zinrai Server Inference, which efficiently performs real-time analysis of high-resolution, high-quality video images, a process that cannot be achieved with conventional cloud video processing systems. Applications that can benefit from the video analysis system include advanced surveillance camera, digital marketing, sports video analysis, and content management in video distribution services. Socionext plans for the delivery of the system in December 2018.

Socionext – www.socionext.com


Vous êtes certain ?

Si vous désactivez les cookies, vous ne pouvez plus naviguer sur le site.

Vous allez être rediriger vers Google.