This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Read our Private policy>

Need Help?

Huawei and SenseTime Jointly Launched the High Capacity Face Recognition Integrated Solution

10/30/2017 12:00:00 AM Author: Source:
[Shenzhen, China, October 30, 2017] At the 16th China Public Security Expo, Huawei and SenseTime jointly launched the high capacity face recognition integrated solution. Powered by Huawei's G2500 smart video analytics server and SenseTime's face recognition algorithm, the solution has been specifically tuned in software and hardware. Each single server supports 224 channels of high-definition video streams for face recognition and real-time processing. The solution delivers new levels of precision, efficiency, and ease of use, adding fuel to smart cities.

华为IT产品线副总裁邱隆(右)和商汤科技副总裁柳钢(左)出席发布仪式。Qiu Long, Vice President of Huawei IT Product Line (Right) and Liu Gang, Vice President of SenseTime (Left) at the launch ceremony

Qiu Long, Vice President of Huawei IT Product Line (Right) and Liu Gang, Vice President of SenseTime (Left) at the launch ceremony

The exponential increase of surveillance cameras and increasingly higher definition of images have posed tremendous challenges on the video processing capabilities of surveillance systems. Over the recent years, the blazing-fast advancements in deep learning technologies have made it possible to handle face recognition, data processing, and analytics in a more efficient, intelligent, and automated manner through artificial intelligence. This is becoming an unstoppable trend. However, currently in China, small- and mid-sized face recognition systems have become a typical difficult scenario in smart video analytics deployment. For example, according to the traditional CPU-based video processing method, to deploy a face recognition system below 200 channels, a large number of servers are required. This means high device costs, large footprints, and complicated maintenance. All these factors have barred wider adoption of face recognition solutions.

The high capacity face recognition integrated solution released this time is a perfect combination of hardware, AI algorithm, and application. On the hardware side, the solution is built on the Huawei G2500 smart video analytics server equipped with 16 NVIDIA Tesla P4 GPUs. The AI algorithm has been deeply optimized by SenseTime for GPU acceleration, which enables speedy execution of tasks such as face recognition, tracking, key point locating, and characteristics extraction. Each G2500 server supports up to 224 channels of high-definition video streams for real-time face recognition and analysis. These benefits especially address the requirements of the small- and mid-sized intelligent video surveillance scenarios for face analysis, security control deployment, snapshot library searching, and picture storage. The solution brings multiple advantages, including high definition, efficiency, and ease of use, and is suitable for industries such as security, transportation, enterprise, and education.

Qiu Long, Vice President of Huawei IT Product Line, said, "Safe City is a key focus of Huawei in the enterprise business. Huawei is glad to team up with SenseTime in video analytics. SenseTime's advanced algorithms and applications in the AI domain have strengthened Huawei's video analytics solutions. I believe that our partnership makes us better positioned to build safer, better cities together."

Liu Gang, Vice President of SenseTime, said, "Intelligent video technology is key to building smart cities. That requires not only superior compute power, but also state-of-the-art intelligent algorithms. These are the prerequisites for making faster, precise video understanding and analysis possible." He added, "SenseTime and Huawei have a long and in-depth partnership in the smart video. Our joint release this time not only presents an algorithm and hardware integrated solution. Rather than that, we bring together our professional insights and understanding in smart video technologies. I think this will pave the way for scale implementation of smart video solutions in city management."


Huawei FusionServer G2500 Smart Video Analytics Server

The Huawei FusionServer G series heterogeneous computing servers are specially engineered for AI services. The G2500 smart video analytics server is positioned for scenarios such as safe city and smart transportation. Each 4U chassis supports 2 CPUs, 16 NVIDIA Tesla P4 GPUs, and 24 3.5-inch hard drives. The configurations enable the server to effortlessly provide inference, analytics, and data storage for 256 channels of HD video streams, making the server ideal for integrated video analytics solutions.