Let's see this wonderful "fingerprint" Paul... Video fingerprinting is used for copyright purposes and is of no use in detecting "suspect" videos. The AI algorithm required to do this would require a lot of processing power. Just how is a provider supposed to finance the development of said algorithm...? And then apply it in real time across an entire network? The computational power required would be enormous, thus YouTube's abject failure in this area.
Open NSFW is an open source neural network that struggles with static images... How is a provider supposed to monitor video in real time? An interesting Open NSFW talk here - https://www.youtube.com/watch?v=02Bmt7tksvM Andy -----Original Message----- From: AusNOG <[email protected]> On Behalf Of Peter Fern Sent: Tuesday, 9 April 2019 2:30 PM To: [email protected] Subject: Re: [AusNOG] More legislative interventions On 9/4/19 2:22 pm, Paul Wilkins wrote: > 2 - Ensure you have in place a mechanism to match electronic > fingerprints of material similar to anything identified in a eSafety > Commissioner's notice. > > By the by, without a mechanism for the eSafety Commissioner to match > content (a common mechanism for electronic fingerprinting material > across hosting providers), the eSafety Commissioner will find > themselves playing whack a mole chasing content specific to each > hosting provider. What do you think that looks like, exactly? You've brought up this magical fingerprint technology multiple times, and been rebuffed multiple times, with no response. I think it's irresponsible to suggest that this is an easy solve. _______________________________________________ AusNOG mailing list [email protected] http://lists.ausnog.net/mailman/listinfo/ausnog --- This email has been checked for viruses by AVG. https://www.avg.com _______________________________________________ AusNOG mailing list [email protected] http://lists.ausnog.net/mailman/listinfo/ausnog
