(Yicai Global) Oct.31 -- Beijing SenseTime Technology Development Co. (SenseTime) and Huawei Technology Co. have jointly unveiled the SenseAtlas ultra high-precision face-recognition solution. Supported by GPU acceleration and SenseTime algorithm, a single server can achieve real-time face recognition via 224-series high definition video streaming, thus offering the best artificial intelligence solution equipped with software and hardware for the smart video analysis cloud market worth 100 billion, people.cn reported today.
SenseAtlas, the integrated machine with the ultra-high-precision face-recognition, combines hardware, AI algorithm and application. The integrated machine applies Huawei Atlas platform technology and combine the SenseTime algorithm to deeply optimize the GPU acceleration, which can complete face detection, tracking, key point positioning, and feature extraction efficiently, replacing the traditional video processing solution based on CPU that is costly, large and complicated to maintain, the report detailed.
Based on its powerful performance, SenseAtlas enables real-time facial recognition via 14-series high definition video streaming on one single GPU, while a maximum of 224 series HD video streaming is allowed on one single server, capable of meeting the demand of small- and medium-sized smart video monitoring scenarios in video face analysis, deployment, snapshot library inquiry and picture storage. Featuring high recognition rate, usability, efficiency and other advantages, this product can be applied in buildings, museums, education, enterprises, transportation, parks and an array of industries and scenarios.
As a key to pushing the development of smart city, the smart video technology requires not only the strong support of algorithm, but also an advanced and intelligent algorithm so as to analyze and understand video faster and more accurately, said Liu Gang, SenseTime vice president.