Anyone in Korea can now use more than 1.17 million broadcast video datasets for AI training in education and research.
The Broadcasting and Media Communications Commission and the Korea Radio Promotion Association said they are releasing more than 1.17 million broadcast video AI training datasets for educational and research use.
The data is being released through the "AI Hub Safe Zone" operated by the National Information Society Agency (NIA). It is divided into 10 major categories, including video and image understanding and description, background, character and object generation, and production environment enhancement. A distinctive feature is that it reflects Korean culture and sentiment, including broadcast video images, backgrounds and characters, objects, Hangul fonts, and subtitles.
Although broadcast video has been evaluated as high-quality data suitable for AI training, its use has been limited due to issues such as intellectual property rights, portrait rights, and neighboring rights. The commission expects that this data release will allow developers and researchers to use the data for research and development of various AI models without the burden of rights issues.
The AI Hub Safe Zone is a service that provides AI training data in a security-controlled environment, and anyone can use it free of charge after going through an application and approval process. "Broadcast video is a core asset of the AI industry that concentrates Korean culture, sentiment, and production technology," said Cheon Ji-hyun, Director General of the Broadcasting and Media Promotion Bureau. "We will continue to build and release high-quality AI training data so that it can be used in various industrial fields."
