Recently, the System on Chip technology provider Socinext has announced its collaboration with Foxconn and Network Optix to avail a new AI solution which is scalable and bears high performance for better results.
The Network Optix is a developer of Video management software (VMS) while the Foxconn is one of the global leaders to provide smart technology solutions.
The newly launched Edge-AI solution powered by 24-core Arm Cortex-A53 SoC with a scalable design. As per the press release, it is meant to compensate the today’s computing demand of smart energy, Internet of Things (IoT), and real-time data processing applications.
Socionext and Foxconn together have built BOXiedge which is a fan-less server, packed in a box of 200mm x 200mm (1U) and typically consumes only 30W of power.
According to the company, the BOXiedge can provide 20TOPS and with the help of an integrated AI accelerating card, it will prove an ideal solution to industries deals in internet AI applications.
Furthermore, it also supports open-source Artificial intelligence development frameworks such as Caffe and TensorFlow, so if the developer already has a knowledge of such AI frameworks then no need of going through learning curves.
In addition to all that for better optimization and to handle the extra load of processing created by real-time applications, Foxconn has pre-installed the Network Optix’s Witness VMS to the BOXiedge server along with immediate support for a broad range of IP cameras on the market.
Also, the launched BOXiedge Plus Nx Witness VMS can be integrated seamlessly with other products ‘Powered-by-Nx’ built on the Nx Meta Video Development Platform for analyzing and enriching video data.
Therefore, the user can process multiple videos in real-time. As it also supports the management of IP-cameras video streams, thus to view and manage them an intuitive graphical user interface will also be supplied.
According to the Socionext, the launched Edge AI server is a great solution for real-time edge inference applications such as being able to recognize and filter video input using metadata to identify objects, people, commodities, human faces and even pathways. Potential applications include smart retail, smart manufacturing, surveillance, medical AI, and more.