Back to the newsroom

Press Release

Korea Deep Learning selected for project to build the world's first AI training data for ink wash painting

Korea Deep LearningPublished

Korea Deep Learning selected for project to build the world's first AI training data for ink wash painting
In September 2022, Korea Deep Learning formed a consortium with Kyung Hee University, Seoul National University and Allbigdat and was selected for a project to build the world's first AI training data for ink wash painting.

In September 2022, Korea Deep Learning was given final selection for the 2022 AI Training Data Construction Project and will build the world's first AI training data for ink wash painting. The work organizes the styles and techniques of ink wash painting, which is expressed with water and ink, into data that AI can learn from, and Kyung Hee University, Seoul National University and Allbigdat are taking part as a consortium.

The "traditional Korean ink wash painting" field of the Data Dam project

The AI Training Data Construction Project is a "Data Dam" project that creates the data needed for AI training on a large scale and opens it to the public through AI Hub. It is overseen by the Ministry of Science and ICT and carried out by the National Information Society Agency (NIA), with the aim of lowering the barriers businesses and institutions face in adopting and developing AI.

In April 2022, Korea Deep Learning was selected for the "Korean traditional ink wash painting data by painting style" field of the project. Until now, no one had organized the expressive techniques of ink wash painting, in which the same subject can be painted completely differently depending on the style and technique, into data that AI can learn from.

From artists' hands to image data

The project is carried out by a consortium made up of Kyung Hee University as the lead organization and Seoul National University, Allbigdat and Korea Deep Learning as participating organizations. The consortium plans to finish building the data by the end of November 2022.

The data is built by the hands of artists and experts. A total of 2,200 original ink wash paintings, reclassified by difficulty, will be distributed to carefully chosen ink wash painters, who will work with experts in each field to turn the many expressive techniques of ink wash painting into image data.

Korea Deep Learning handles data processing and model development

Within the consortium, Korea Deep Learning is responsible for data processing and model development. Korea Deep Learning has a record of deploying AI solutions at public institutions such as the Korea Centers for Disease Control and Prevention and at large financial companies such as Shinhan Financial Group, and it holds six AI patents. Building on this experience, it is expected to deliver results in AI model development as well as data processing.

"Through this consortium, we plan to turn the many expressive techniques of ink wash painting into data in a form that AI models can easily use," said Kim Ji-hyun, CEO of Korea Deep Learning. "The ultimate goal of this project is to use AI technology to inspire new directions in ink wash painting education and to strengthen interest in and understanding of traditional art."

KB Unicorn Club and logistics OCR

In July 2022, Korea Deep Learning was selected for the second cohort of KB Unicorn Club, a joint accelerator program of the Gyeonggi Center for Creative Economy and Innovation and KB Kookmin Bank. The program finds promising startups less than three years old and supports them with IR capability training, customized advancement programs, and investment review through a demo day. Korea Deep Learning was recognized for the technology behind its core product, DEEP OCR, and also received the 2022 Korea Impactech Award.

Around the same time, Korea Deep Learning introduced an "ultra-high-speed deep learning-based logistics waybill recognition system" that immediately re-recognizes waybills when a barcode is misread, reducing unsorted items. Selected as an outstanding logistics technology company and an incubated company, it ran a pilot project with Korea Post, and in 2021 it signed a consortium agreement with Sunghwa Corporate Express, the No. 1 company in B2B pouch delivery.

About Korea Deep Learning

Korea Deep Learning (KDL), founded in 2019, is an enterprise AI agent company. Built on KDL Frontier, its own vision-language model (VLM) specialized for documents, KDL makes AI agents that take on and complete the document work of businesses and public institutions. Its main products are DEEP OCR, DEEP Parser, DEEP Agent and DEEP Ops.

Source