boduzapho |
07-12-2010 06:40 AM |
I would have to say monitoring tube sites for stolen video would be a massive undertaking as automated scanning would be next to imposable.
Tube site rename videos , change the size etc.. pattern matching which is what a spider would have to do based on those basic properties would fail.
I, myself, and a team of guys from the U of Chicago, in the 90's developed image recognition software to detect colors, facial features etc... This method could work but I expect the hardware, software costs would be enormous, not to mention the bandwidth bills to monitor these tube sites for content. (You may remember Lycos who licensed this software, they were the first to AI images, so no it was not Google or Yahoo)
I have created algorithms that can be used to detect duplicate images, so grabbing a screen shot of a video and comparing it to what you have is possible but again you would have to process the entire video frame by frame.
Another option which would have a very large overhead would be do download the new daily videos or log the URL's and spend the day looking for your stuff with human eyes. Again not very effective.
So no offense guy but I doubt you could offer up a reasonable solution on a programming level that would be of any real use.
|