Videorooter: technology to uniquely identify video

Images and videos travel from platform to platform on the Internet. In this process the provenance information of these media files, that is information on for example the maker and the terms and conditions under which the work can be shared or reused, often gets lost. This makes it difficult to trace who is the original owner of the work. Even if extensive license information++Creative CommonsFor example: Creative Commons licenses give copyright holders the possibility to share their work under conditions other than all rights reserved. KL is the public lead of Creative Commons Netherlands. More information has been provided in the initial publication, there is no guarantee this stays with the image when the work is shared on third platforms. Videorooter is a first iteration to solve this issue for online video.

Kennisland’s work in the heritage sector has made clear that ownership on the Internet is a complex issue, and thus far no good solution has been offered. We find it particularly important to find a solution for makers who release their work under an open license. They deserve proper attribution when their work is shared or reused. Kennisland designed Videorooter in the hope to strengthen the position of makers who share their videos under open licenses.

Kennisland and the Swedish Commons Machinery develop fingerprinting technology to uniquely identify videos based on specific characteristics. This technologyOur work in the heritage sector has made clear that ownership on the Internet is a complex issue, and thus far no good solution has been offered. can distinguish one video from another to thereby enable the possibility to link provenance information to a video. This also makes it possible to trace if one video is posted on multiple platforms.

A unique code per video

In the first phase of the project, Commons Machinery develops algorithms to fingerprint videos. They built on knowledge and experiences from the past. In their project elog.io Commons Machinery has already developed fingerprinting technology for images. They now expand this technology for videos.

Fingerprinting++FingerprintingThe algorithms are designed such that they analyse videos based on traits that can be distinguished by the human eye. So if a person would say that two videos are one and the same, they would be fingerprinted with the same code. means that one video gets assigned a specific code. This code is not only unique but also specific. This means that as long as the same algorithm is used, the same code will always be assigned to a certain video. This applies even when the video has gone through small modifications such as a change in resolution or format. The algorithms can also identify derivative works++Derivative worksA derivative work is one that comes about by mixing, adjusting and adding features to an existing work., and the amount of similarity between the derivative and the original video.

After the algorithms have been developed, we will fingerprint 50,000 openly licensed videos from the platforms Europeana, Wikimedia Commons and Open Images++Technological innovationKL has played a role in the development of all three platforms. Read our case on technological innovation in the cultural heritage sector for more information.. We then store the unique codes that come out of this process, together with other metadata such as title, maker and license, on a whitelist.

Simultaneously we translate the algorithms in open standards for video fingerprinting. Having clearly definedWe have designed Videorooter in the hope to strengthen the position of makers who share their videos under open licenses. open standards simplifies the process of implementation of our algorithms. This makes it easier for others to use our technology. The standards are defined such that anyone could implement the algorithms, regardless of the used programming language.

Towards a strong and open Internet

Videorooter is part of a bigger solution that Kennisland designs for the problem of ownership on the Internet. In the overarching project Octopus++OctopusChair of PeerPractice Catharina Maracke explains the design of Octopus in more detail here., Kennisland and PeerPractice develop a system that regulates ownership and attribution on the Internet. Piece by piece, we hope to strengthen the position of makers who make their creative works available under open licenses, to thereby create an Internet on which makers dare to share.

Videorooter is made possible with the support of the Dutch SIDN fund.

• KL'ers involved
• Tessa Askamp tas@kl.nl
• Maarten Zeinstra mz@kl.nl