AI to prevent video content illegal leakage…Research to be started in the next month
Lee Gyung Tak | kt87@ | 2017-07-24 10:40:54
#Movie ‘Okja’ of Bong Joon-ho has been leaked after it was revealed in Netflix in 12 hours. The movie ‘Real,’ which was the subject of the actor Sully`s film, also went online on the day of its release, and a cable TV movie called "King of the American drama" It is spreading indiscriminately through eTorrent, WebHard, P2P, and others.
A study on the use of artificial intelligence (AI) to fundamentally prevent the illegal outflow of video content such as the latest movies and dramas is getting worse. The Korea Copyright Supervisory Service (KIPO) will start a research project on ‘AI Usage for Prevention of Copyright Infringement of Video Material’ from next month, and will design an algorithm model by the end of this year and plan to build a dedicated server system equipped with a GPU (graphics processing device) from next year.
Looking at the copyright protection annual report of the protector, about 80% of the video contents in the damage of piracy and outflow in the domestic contents market are occupied. Video content distributors such as Netflix have introduced digital rights management (DRM) solutions to their platforms for copyright protection, but hackers are using it as an advanced technology to disable it. The public can easily find hacking tools that can disable DRM through portal search.
The related personnel had to manually take measures such as building a minutiae DB for each image in order to directly identify and block the leaked data. Especially, it is the protector`s explanation that the existing feature point technology was difficult to identify the image quality change or small data change of the same image.
The protector is trying to introduce AI technology in order to overcome the technical limitations of existing feature point based filtering technology. Designing a research model based on deep learning to protect the movie from film or distributors Even if the movie is illegally circulated in online form, not only full version but also various pieces of video material such as sculpture, it is aimed to identify the video object.
Even if the guard has not secured the original image of the movie, such as an unopened movie in order to identify pirated images, the company plans to use an image of a trailer or a movie poster to identify an illegal video with an accuracy of more than 95% to be. An official from the Copyright Protection Protection Team stated, "We will select the companies that will perform the next week and conduct basic research until the end of this year and conduct follow-up business once the budget is secured through validation process. It will be used for the inside of the guard.”
By Lee Gyung Tak kt87@
[ copyright ⓒ The Digitaltimes ]
A study on the use of artificial intelligence (AI) to fundamentally prevent the illegal outflow of video content such as the latest movies and dramas is getting worse. The Korea Copyright Supervisory Service (KIPO) will start a research project on ‘AI Usage for Prevention of Copyright Infringement of Video Material’ from next month, and will design an algorithm model by the end of this year and plan to build a dedicated server system equipped with a GPU (graphics processing device) from next year.
Looking at the copyright protection annual report of the protector, about 80% of the video contents in the damage of piracy and outflow in the domestic contents market are occupied. Video content distributors such as Netflix have introduced digital rights management (DRM) solutions to their platforms for copyright protection, but hackers are using it as an advanced technology to disable it. The public can easily find hacking tools that can disable DRM through portal search.
The related personnel had to manually take measures such as building a minutiae DB for each image in order to directly identify and block the leaked data. Especially, it is the protector`s explanation that the existing feature point technology was difficult to identify the image quality change or small data change of the same image.
The protector is trying to introduce AI technology in order to overcome the technical limitations of existing feature point based filtering technology. Designing a research model based on deep learning to protect the movie from film or distributors Even if the movie is illegally circulated in online form, not only full version but also various pieces of video material such as sculpture, it is aimed to identify the video object.
Even if the guard has not secured the original image of the movie, such as an unopened movie in order to identify pirated images, the company plans to use an image of a trailer or a movie poster to identify an illegal video with an accuracy of more than 95% to be. An official from the Copyright Protection Protection Team stated, "We will select the companies that will perform the next week and conduct basic research until the end of this year and conduct follow-up business once the budget is secured through validation process. It will be used for the inside of the guard.”
By Lee Gyung Tak kt87@