The Silicon Review
“EDGEMARIX is widely used in the fields of social infrastructures, such as roads, parking lots, railroad tracks, stations, large-scale facilities, and long-term care. I have come to the conclusion that our expertise is important for realizing a safe and secure society.”
In Japan, the number of network cameras installed is smaller than in Europe and the United States, and the analysis and utilization of video by AI is delayed. As a leader company in ‘Video Edge AI’, EDGEMATRIX believes that it is important to increase the number of adoption cases and widely convey its merits.
In addition, the installation environment at the site varies and is affected by the weather and time. Therefore, the setting for implementing ‘Video Edge AI’ in the field and obtaining high recognition accuracy becomes complicated. EDGEMATRIX is packaged according to the application so that this installation can be made as easy as possible. Then, semi-customization is performed so that the user's request can be easily met.
EDGEMATRIX develops fundamental technology of edge AI for processing real-time video data and provides it as a platform service. Currently, for example, the company is developing ‘Video Edge AI’ services in the field of long-term care — to predict the elderly from falling. This makes it easy to introduce ‘Video Edge AI’ in the healthcare and long-term care fields.
EDGEMATRIX provides the infrastructure and platform for implementing ‘Video Edge AI’ in the field, rather than developing AI.
‘Video Edge AI’ processes surveillance camera video footage using AI at the installation site (edge) and utilizes it in real-time. ‘Video edge AI’ is a new business field that is different from conventional network cameras and cloud AI. The advantage is that you don't have to send camera footage to the cloud over the network. AI can warn on-site dangers and abnormalities in real-time and take action. And since there is no need for video compression for transmission, AI can analyze high-definition video and the recognition accuracy is high.
The Silicon Review reached out to Hiroshi Ohta, CEO and co-founder of EDGEMATRIX, for his comments about the application of ‘Video Edge AI’. Here’s what he said.
“EDGEMARIX is widely used in the fields of social infrastructures, such as roads, parking lots, railroad tracks, stations, large-scale facilities, and long-term care. I have come to the conclusion that our expertise is important for realizing a safe and secure society. So, it is required to continue to provide high-quality services for a long period while responding to the ever-increasing number of installations and the number of subscribers. Currently, it is a startup with about 30 employees, but it will be an important challenge to build and maintain a corporate structure that can realize this in the next five years.”
‘EDGEMATRIX Store’ — An Effective Facilitator
To realize a smart city, many various AI apps are required to recognize people, objects, and cars from camera images. The unique point of EDGEMATRIX is that instead of developing this huge number of AI apps in-house, it operates the ‘EDGEMATRIX Store’ and introduces and sells AI applications developed by developers and partners. In other words, by providing the infrastructure and platform for implementing AI apps in the field and operating the store of AI apps that operate there, the ecosystem of ‘Video Edge AI’ is formed.
This is the same as the relationship between smartphones such as the iPhone and the App Store that provides the app. Usually, it charges a lot of development costs to introduce AI. For users, the AI app from the EDGEMATRIX store has the advantage that it can be used for about $100 per month. For AI developers, selling at the store allows them to sell their development results “n” times.
Infrastructure for Implementing ‘Video Edge AI’
‘Edge AI Box’: It’s a compact device equipped with a GPU for AI processing of video in the field, a communication function for transmitting analysis results and video, and various interfaces for connecting to cameras and displays.
“Many cases where surveillance cameras are installed are outdoors, and the ‘Edge AI Box’ is installed right next to it, so robustness is required. Therefore, we have a rich lineup not only for indoor use, but also for outdoor installation with dustproof, waterproof, and lightning strike measures, and 5G support. Although the risk of theft is high outdoors, strict security measures are taken to prevent theft of hardware, software, and stored video data.”
‘EDGEMATRIX Service’: It’s a platform service that provides the three Rs (Remote, Realtime, and Resilience) required for the on-site implementation of ‘Video Edge AI’.
Remote: It is impossible to manually manage thousands of devices installed in Japan and overseas, and a function that can automatically manage them remotely is indispensable. The ‘EDGEMATRIX Service’ can remotely manage cameras, Edge AI Boxes, and AI apps.
Real time: When danger or anomaly occurs in the field and AI warns, it is necessary to be able to monitor the field image in real-time and judge the situation. At that time, many people must be able to watch many images at the same time from anywhere via the Internet. Since the ‘EDGEMATRIX Service’ transmits video by P2P without going through the cloud, it is possible to monitor the video in real time.
Resiliency: In addition to fault tolerance, there must be a mechanism in place to prevent unauthorized use and hacking. The ‘EDGEMATRIX Service’ realizes Resilience with zero trust architecture.
‘Edge AI Solution’: It provides consulting, customization, and optimization of AI applications and supports installation work.
“In the future, a huge number of network cameras will be installed as the definition of images increases. It will be difficult to transmit all the videos over the network and centrally process and store everything in the cloud. Therefore, it is inevitable that the videos will be processed by AI at the edge and utilized, and there is no doubt that society will need the infrastructure and platform to realize this. That said, we manage our company under a corporate culture where it is natural to change plans flexibly and respond quickly without being bound by a fixed plan.”
Hiroshi Ohta | CEO & Co-Founder
Hiroshi Ohta was the development manager for a Japanese mobile telecommunications company, providing the world's first mobile phone camera-equipped products and services. After serving as an executive of Vodafone Japan and Softbank Mobile, he founded Cloudian Inc, headquartered in Silicon Valley, with a friend and served as CEO of its Japanese entity. In 2019, he founded EDGEMATRIX, a startup spun out from the AI project team of a global venture company.