EdgeTuring AI Chip Seeks to Improve Video Analytics
March 3, 2021
A collaboration between Leopard Imaging, Hailo, Socionext and Amazon Web Services has produced EdgeTuring, an artificial intelligence processing solution for video analytics. The venture aims to produce high image quality and energy efficiency for AI inference nodes, benefiting several applications in industrial automation, smart devices, and smart retail, among others.
Leopard Imaging, which develops embedded cameras, said it has been working to address the need for affordable multiprocessing power in deep learning applications. Using Socionext’s SC2000 image signal processor and the Hailo-8 M.2 AI acceleration module, the EdgeTuring consumes less power, performs at a higher level, and provides greater reliability for video analytics and privacy at the edge than alternative solutions, Leopard Imaging said. By leveraging AWS services such as Amazon Kinesis Video Streams and Amazon Kinesis Data Streams (Amazon KDS), EdgeTuring can create “a seamless experience for customers to stream and analyze videos using a simple Internet connection.”
The EdgeTuring has an accuracy that ranges from 95% to 99% for several deep-learning-based computer vision applications, including object detection, image classification, and others, Leopard Imaging said. “It processes frames much faster, supports more functions, consumes less power, and costs much less than any comparable solution – all with the ability to stream inputs in real time,” said Bill Pu, president and co-founder of Leopard Imaging. “We believe that this strategic collaboration will help us carve a new path forward in the AI-driven camera industry. We want to break away from the status quo and embrace these opportunities to adopt new AI solutions.”
Launched last year, the Hailo-8 enables customers to integrate high-performance AI capabilities of 26 Tera Operations Per Second (TOPS) into edge devices, providing more flexible options to accelerate deep learning-based applications within a standard form factor, the company added. Comparing Hailo-8’s average frames per second with competitors across multiple standard NN benchmark showed that Hailo’s AI modules achieve up to 26x higher FPS rates.
“We are excited to work with Leopard Imaging, Socionext, and AWS to bring unprecedented AI-based products to the edge market,” said Orr Danon, CEO and co-founder of Hailo. “As demand for AI at the edge grows, our innovative AI processor and acceleration modules, together with the solutions created by these industry leaders, will help usher in a new generation of AI that is more powerful, more scalable, and more cost-efficient.”