More on Philips’ anti-piracy video tech


As we reported last week, Philips Electronics says it has developed technology that will use digital fingerprinting to help online video sites restrict use of copyrighted content. Since then we caught up with Ronald Maandonks, CEO of Philips’ content identification division, for more details in a brief phone call.

While he wasn’t long on specifics of the technology or the business plan, Maandonks said the MediaHedge technology works by taking a bitstream “fingerprint” from a video clip to create a database of known material; services using the MediaHedge system, he said, will then be able to compare the database fingerprints to videos being uploaded, and can block or filter those that might be violating copyrights.

“The big challenge is building something that is fast enough, and accurate enough,” said Maandonks, who claims that MediaHedge, upon release, will be fast and accurate enough to let video-sharing services match and restrict videos while they are being uploaded. According to Maandonks, the MediaHedge fingerprint “is very small — for 5 seconds of material, the fingerprint is just a few kilo[bits].” The system, he said, uses a Linux OS and runs on “off-the-shelf” PC hardware.

Many questions remain, such as who will build a video database, and how will that be done — and whether or not Philips will license the technology, sell it, or offer it as a service. While Maandonks would not discuss MediaHedge business models (“now is not the time”), he did say that Philips “is in contact” with several potential clients. Philips plans to demonstrate the MediaHedge technology at the upcoming CES show in Las Vegas.

There are precedents here. Philips has an existing contract with Snocap for parallel technology that does audio fingerprinting for music. And Snocap rival Audible Magic recently announced it had licensed video fingerprinting technology from an independent scientist. Video aggregator Guba has an internally developed video fingerprinting system, which CEO Tom McInerney told us he discussed licensing to YouTube and other site. However, YouTube has since said it is working on similar technology on its own.

While the market is certainly there — just ask sites like Bolt or Dailymotion, which are either already being sued or fear legal repercussions for unlicensed content — other challenges await. Once you’ve figured out the technology, there’s the problem of scaling it, and also the task of brokering deals to get content to input into the system.

Update: Another company is launching in this space, Attributor, which has $10 million in VC funding. There’s a profile in the WSJ today. We’ll have more coverage when we talk to company directly.

With additional reporting by Liz Gannes.


Saurabh Kaushik

These technologies have to be very adaptive in nature as digital media content dynamic changes everyday. Few other companies are trying to develop a business system around these. Philips has to be faster to market.

Chris D

Identification of video is a very, very hard problem to solve, so I’d be a bit skeptical of these solutions. This is a type of technology that people go do PhD’s and post-Doctorates on at MIT and Standford just to do some limited solutions.

They may be able to detect copyrighted material in some specialized cases, but are most likely not robust in the general case. However, I’m sure that Services that license these semi-solutions would at least get some “in good faith” recognition from Content Owners.

Comments are closed.