In short, yes. The key innovation is that we've come up with a lossy, fixed-length representation of the visual features that we can use to do classification up

In short, yes. The key innovation is that we've come up with a lossy, fixed-length representation of the visual features that we can use to do classification upon. I'll try to do a more detailed writeup on our blog when I find some time. btw, our blog has no rss feed, but you can just use our RSS API :-) http://www.diffbot.com/api/rss/http:/www.diffbot.com/blog http://www.diffbot.com/api/rss/http:/www.diffbot.com/blog