Mobile processor maker Qualcomm announced on April 12 that it will launch the Qualcomm Vision Intelligence Platform. The platform is equipped with the first system-on-chip (SoC) series specifically designed for the Internet of Things (IoT) using the advanced 10nm FinFET process technology - QCS605 and QCS603. The two system single chips can be used for a wide range of IoT applications to provide powerful computing capabilities for the camera's built-in camera processing and machine learning, as well as excellent efficiency and cooling efficiency.
Qualcomm pointed out that the QCS605 and QCS603 systems integrate Qualcomm's advanced image signal processor (ISP) and Qualcomm artificial intelligence (AI) engine on a single chip, and also include ARM's advanced multi-core CPUs, vector processors, and GPUs. The vision intelligence platform also includes Qualcomm's advanced camera processing software, machine learning and computer vision software development kit (SDK), and Qualcomm's proven connectivity and security technologies. The platform has been optimized to bring exciting new opportunities and possibilities for industrial and consumer smart security ip camera, motion camera, wearable camera, virtual reality (VR) 360-degree and 180-degree camera, robots and smart screen. Currently including KEDACOM and Ricoh THETA are all planning to develop products based on Qualcomm's visual intelligence platform.
Qualcomm further emphasized that the visual intelligence platform integrates a Qualcomm artificial intelligence engine (AI Engine) composed of multiple hardware and software components to accelerate terminal equipment artificial intelligence. Qualcomm artificial intelligence engine includes Qualcomm Snapdragon Neural Processing Engine (NPE) software framework (for development with the TensorFlow, Caffe, and Caffe2 frameworks, tools for analysis, optimization, and debugging), Open Neural Network Exchange (ONNX) exchange format, and Android Neural Network API and Qualcomm Hexagon Neural Network Library. All of the above designs are designed to support developers and OEM vendors so that they can easily connect training networks to this platform. Moreover, through the Qualcomm artificial intelligence engine and the Snapdragon neural processing engine software framework, the visual intelligence platform can provide up to 2.1 trillion operations per second for deep neural network reasoning, improving performance over other leading alternative solutions.
没有评论:
发表评论