ETRI Unveils No-Code AI Tools to Boost Industrial Applications

The Electronics and Telecommunications Research Institute (ETRI) has launched a suite of no-code machine learning tools designed to facilitate AI development for users with limited technical expertise. This release, which includes the core technology of the TANGO framework, allows for the automatic generation of neural networks and streamlines deployment processes. The announcement came during a public seminar held on November 6, 2023, at the Science and Technology Center in Gangnam-gu, Seoul.

The TANGO framework is engineered to simplify the development of application software that integrates artificial intelligence. It effectively deploys applications across various hardware environments, including cloud infrastructures, on-premise setups, and mobile devices. This innovation addresses challenges in industries such as manufacturing and healthcare, where the application of AI has often been stymied by technical barriers. For instance, while quality inspections in steel mills can easily identify defects, leveraging AI for such tasks has proven complex. Similarly, doctors can diagnose conditions like tuberculosis through X-ray images, but automating this process with AI has been problematic.

ETRI’s TANGO framework is particularly beneficial for domain experts lacking deep knowledge of neural networks. It can be installed with a simple command and accessed immediately via a web interface, making it user-friendly. Traditionally, the development of AI applications has required domain experts to manage data labeling while software developers handled model training and deployment. The growing demand for AI solutions in various sectors has highlighted the shortage of skilled AI and software professionals.

To mitigate these challenges, ETRI has developed a neural network automation algorithm tailored for object recognition, addressing the needs of local industries. The organization also introduced LLMOps tools to facilitate the development of generative AI applications. The TANGO project has resulted in significant achievements, including the filing of 24 patents, publication of 3 NeurIPS papers and 13 SCI papers, and over KRW 10 billion in commercialization revenue.

One notable success is the collaboration with Avenotics, a company specializing in autonomous maritime navigation solutions. Avenotics was selected for the “Public Research Outcome Expansion and Commercialization Project” by the Ministry of Science and ICT, aiming to commercialize promising research outcomes from government-funded institutes. The company secured an investment of 1.3 billion won from various stakeholders, including Korea Science and Technology Holdings and the Korea Credit Guarantee Fund. Through this partnership, Avenotics is utilizing TANGO’s technology to enhance on-device AI capabilities, which automatically generates contextual navigation information.

ETRI’s research team is actively promoting wider adoption of the TANGO framework by conducting pilot demonstrations in collaboration with partner institutions. For example, Weda Co., Ltd. has developed an AI service for on-site employees in the steel and automotive parts sectors, focusing on vision-based inspections of complex shapes. Additionally, Lablup Inc. has partnered with KT Cloud to launch a deployment optimization service supporting the latest AI acceleration engine, ATOM-Max, and has commercialized a GPU cloud rental service in cooperation with KT.

In the healthcare sector, Seoul National University Hospital is advancing AI technology to analyze large-scale chest CT images and diagnostic data. This initiative aims to automatically generate diagnostic reports, with plans for evaluation and verification at multiple hospitals, including Seoul National University Bundang Hospital and Boramae Hospital.

The LLMOps tool, which supports generative AI development, is also progressing toward commercialization through a partnership with Acryl Inc., with the source code of Acryl’s product, Jonathan, fully available on GitHub.

According to Kim Tae-ho, Software PM of the Institute of Information & Communications Technology Planning & Evaluation (IITP), “TANGO technology is truly the best open source project in Korea and is contributing greatly to enhancing the competitiveness of the domestic software industry in the field of artificial intelligence development tools.” Jo Chang Sik, Principal Researcher at ETRI, indicated plans to expand the TANGO project into LLMOps tools for generative AI, emphasizing ongoing commitment to sharing development expertise.

The research team intends to continue releasing updates on GitHub every six months, alongside annual public seminars to disseminate technological advancements and practical insights. Over the course of four seminars, 944 participants from 552 institutions engaged in discussions about AI technology, illustrating a robust interest in the evolving landscape of AI applications.