Recently, Kreon, a start-up focusing on edge AI, has launched its latest customized chip, the kneron KL 720 SOC. According to the company, in the mobilenetv2 image recognition benchmark, the KL 720 is twice as energy efficient as Intel’s latest mobile chip and four times as efficient as Google’s core edge TPU. < p > < p > edge computing refers to an open platform that integrates the core capabilities of network, computing, storage and application at the side near the object or data source to provide the nearest end service. Its characteristics are low latency, high efficiency, and improved data security. < p > < p > AI edge processing focuses on moving the reasoning part of AI workflow to the device, thus limiting the data to the device. Manufacturers increasingly need more efficient algorithms and hardware to deploy machine learning on the device side, while knoron, headquartered in San Diego and founded in 2015, develops a variety of hardware and software products for intelligent devices to run and drive AI applications. < / P > < p > kneron’s founder and chief executive officer, Albert Liu, received a doctorate in electrical engineering from UCLA. He once held research and management positions at Qualcomm. His team developed nine patents and won the Qualcomm impaqt R & D award. He also worked in Samsung Electronics R & D center and worked as a researcher at MSTAR and wireless information. The products of

Kneron include AI chip and other hardware and artificial intelligence models, which can be applied to many kinds of Internet of things devices such as self driving cars, intelligent refrigerators, and so on. < / P > < p > one of the most important products is the knoo edge AI open platform, which uses Kreon driven devices to build grid intelligence. This platform allows collective AI processing and inference across multiple devices locally without using the cloud. The platform is an artificial intelligence ecosystem, open to AI application developers, consumers, SMEs and service providers. < / P > < p > at present, the knoo platform has been applied in Colleges and universities around the world. Through the textbook co authored by Robert Liu, chief executive officer of knoron, and the kneon driven AI module accelerator (named kneo stem), students learn to adopt basic artificial intelligence models and create artificial intelligence applications that can be uploaded to the kneo AI app store. Universities that have joined include UCLA, Notre Dame and the University of Virginia. < / P > < p > in application scenarios, one can connect one knoo stem to the camera and the other to the door lock. When people enter the house, the knoo stem on the door lock will identify whether the entrant has authorization, and the knoo stem on the camera will identify the number of people entering, which can prevent the malicious people from following into the house. < / P > < p > kneo can also connect health monitoring AI application and fitness AI application, monitor and manage doctors’ prescriptions, and report to doctors regularly as a solution for post consultation health care. < / P > < p > last year, the company released the kl520 chip, a chip (SOC) for deploying artificial intelligence on smart home and Internet of things devices. A series of aaeon edge AI modules realize simple edge AI integration. < / P > < p > each module supports different standard PCB interfaces and is designed to accelerate general AI models, such as face and object recognition, gesture detection, etc. This series of edge AI accelerators are developed in cooperation with aaeon. Founded in 1992, aaeon is one of the leading designers and manufacturers of professional Internet of things intelligent solutions. < / P > < p > the company’s latest kl720 chip can process 4K still images and video with a resolution of 1080p. It also brings many new audio recognition breakthroughs for the company, which will allow devices using its chips to bypass standard wake-up words on other chips and have instant conversations with devices. < / P > < p > overall, kneron is committed to a 1.5 tops SOC that uses arm cortex M4 as the main control unit. The average power consumption of the whole package is about 1.2W. “Kl720 opens a new era of intelligent devices by combining unparalleled energy efficiency with kneron’s industry-leading artificial intelligence algorithms,” said Albert Liu, founder of kneron < / P > < p > “its low cost enables more devices to take advantage of the edge of artificial intelligence to protect user privacy to a degree unmatched by competitors. Combined with our existing kl520, we are proud to provide the most comprehensive set of AI chips and software packages for devices on the market. ” < / P > < p > by working with kneo, the company also offers an interesting network solution for devices using its chips. Developers can create their own private networks and connect multiple sensors without routing data to the cloud. < / P > < p > kneron wants to create a market that allows consumers to exchange or sell data to buyers. To ensure security, the network uses blockchain technology to protect data. < / P > < p > “we have a reconfigurable AI engine that supports audio, video, 2D and 3D at the same time, and we’re not just a chip company. We are a company that offers a full range of solutions and can offer a mature model, so it’s almost like a turnkey solution, “Albert Liu said in an interview. < / P > < p > “the purpose of kneron is to spread AI everywhere and personalize it for everyone. Too many artificial intelligence in the world is built in the way of technology first and consumer second, which brings privacy and security problems to consumers and accessibility problems to device manufacturers. Knoron’s mission is to change that and lead to real AI applications in the future. ” < / P > < p > in addition to deploying artificial intelligence at the forefront, kneron has also been on the news for using masks to trick face recognition algorithms. In 2019, they claimed that the 3D mask could crack Alipay and WeChat’s “brush payment” function, and even successfully verified boarding on the mobile phone photos through face recognition at the largest airport in Holland. However, at present, the competition in the field of edge AI hardware is heating up. Another Israeli start-up, hailo, for example, recently released its latest chip. Hailo’s professional hailo-8 Gamma The deep learning processor can perform 26 Tera operations per second (up to the highest), and its innovative architecture enables edge devices to run complex deep learning applications that previously could only run on the cloud. < p > < p > hailo has recently announced the establishment of a subsidiary in Tokyo, Japan, and Hiro Uchida, a former Sony executive, has been appointed president to lead the business. < / P > < p > hailo has also established important partnerships with socionext, a leading supplier of advanced SOC solutions for video and imaging systems, and Foxconn, a global leader in intelligent manufacturing, to launch the next generation of AI solutions for video analysis. [image] Google secretly tests 6GHz networks in 17 states of the United States