VOL.55
August 2019
Special — ETRI to release AI achievements at Germany IFA 2019
ETRI to release AI
achievements at Germany
IFA 2019
A domestic research team will participate in the international home appliance exhibition in Berlin (IFA, Internationale Funkausstellung) and release research achievements related to artificial intelligence (AI). ETRI announced that they will carry out technology marketing by releasing their latest research achievements at the IFA exhibition to be held from September 6. The researchers will showcase four technologies at the exhibition: smart factory operation control system; Internet of media technology; deep learning high-speed processing system; and AI motion analysis technology. In addition, Dr. Kim Hyun of the Smart ICT Convergence Research Department will present a keynote presentation titled “ICT-enabled smart factory” for the world audience at the IFA NEXT Innovation Engine.
Four technologies including smart factory control technology
The first technology to be introduced by ETRI is the smart factory operation control system, which can be used to operate the smart factory for personalized manufacturing based on 3D printers or robots. In particular, it has the benefit of easily changing and reconfiguring the manufacturing line using modular design. In addition, it can manage the smart factory in real time or manage or analyze the operational data using Internet of Things (IoT) sensors. This technology can be applied to the sector that requires customized flexible manufacturing, and experimental products can be quickly produced for small and medium-sized businesses.
Internet of Media Technology mechanism
As the second technology, Internet of media technology verifies whether the media is reliable by analyzing harmfulness, unlawfulness, and correlation among media through the collection of information from various media such as photos and videos using artificial intelligence. Therefore, this technology can be used for the intelligent media information search service and harmful image blocking service. In particular, the research team developed last July a system that deletes illegal videos using this technology. The applied technology automatically detects the same video by comparing the reported illegal video and web hard posts using AI. In the future, the technology can be used to create sound, reliable media services by blocking harmful information.
Deep Learning Acceleration system
The technology to be introduced next is deep learning high-speed system processing technology. A huge amount of data is essential to increase the accuracy of deep learning. Specializing in deep learning, this distributed computing platform technology enables several computers to learn a large-scale deep learning model at high speed when learning a large amount of data such as image or voice. Faster deep learning performance than the existing distributed environment is provided through the optimization of hardware and software for distributed deep learning to overcome the performance limit of deep learning by designing hardware and software together. A company that has constructed a data center already can develop a high-speed deep learning system using this technology.
AI Motion Analysis mechanism
The last technology to be introduced is AI motion analysis technology, which recognizes user’s actions and behavior. Aside from analyzing dynamic and complex movements such as K-POP dances, this technology can also be used for effective posture correction by comparing/analyzing the expert’s posture and learner’s posture and providing the evaluation result. Motion-based educational contents, sports posture analysis and correction systems, and medical rehabilitation clinic service can be created using this technology.
“We plan to carry out technology commercialization marketing overseas and concentrate on technology advertising by releasing AI research achievements — which are implemented by ETRI — to foreign countries by participating in IFA 2019,” said ETRI President Myung-joon Kim.