South Korea Deep Learning Named to Ministry’s “M.AX Specialized Company” Program, Demonstrating Manufacturing Document Datafication Capabilities

South Korea Deep Learning was selected for the “AI solution company” category among the “Manufacturing AI Transformation (Manufacturing AX, M.AX) specialized companies” promoted by the Ministry of Trade, Industry and Energy and the Korea Evaluation Institute of Industrial Technology (KEIT). The selection was announced on the 13th. The M.AX specialized company program aims to discover and nurture specialized companies that can support the manufacturing industry’s AI transformation.

South Korea Deep Learning was included in the category in recognition of its technology for converting manufacturing unstructured documents such as drawings, bills of materials (BOM), and process manuals into data that can be used with AI. Major data categories on manufacturing sites include equipment and sensor data, design information, material composition, process standards, and work methods.

Examples of data accumulated in document form include drawings, BOM, process manuals, and cost estimates. Such document-based data has the limitation that it is difficult to use immediately in AI, Manufacturing Execution System (MES), and ERP systems.

The four fields in which South Korea Deep Learning was recognized for its expertise this time are quality control and inspection automation, generative AI, manufacturing data processing and generation, and manufacturing tacit-knowledge data processing and generation.

South Korea Deep Learning is focusing on converting unstructured documents into structured data. Structured data is intended for use in AI and business systems. To this end, the company has introduced “DEEP Agent for Manufacturing,” a manufacturing-specific document AI solution.

The solution is based on VLM document understanding technology and can identify a document’s structure and context while extracting and structuring the necessary information. Target documents include drawings, BOM, process manuals, and cost estimates, and the system is designed to convert the necessary information in documents with different formats into data. After processing, users can verify the original text and the results, and after linking it with existing business systems such as MES and ERP, it can be used for yield calculations and process data verification.

South Korea Deep Learning has document recognition, structuring, search, and verification technologies, and is envisioning the conversion of process standards, work methods, and expert experience contained in manufacturing documents into data. The key lies in using manufacturing knowledge accumulated in documents in actual operations based on AI.

The company plans to expand its business into the manufacturing sector on the back of this selection. It also recently launched DEEP Ops, a document AI operations platform. Through DEEP Ops, the company aims to expand its scope of support from data build-out to the operations stage.

The platform supports customers in adjusting their own extraction criteria when document formats change. It also supports customers in adjusting their own extraction criteria when work standards change. In addition, it supports customer performance comparisons and verification. The company said it plans to establish a system linking document data build-out, operations, and management based on this.

Kim Ji-hyun, CEO of South Korea Deep Learning, said that the experience and standards from manufacturing sites are accumulated in drawings and manuals. He added that the company plans to convert them into data that can be used with AI. He also said it plans to expand its role from document reading technology to manufacturing data generation infrastructure.

Source: TECHWORLD · Kim Seung-gi
Original: https://www.epnc.co.kr/news/articleView.html?idxno=408104

References

• https://www.epnc.co.kr/news/articleView.html?idxno=408104