Captchas, the bot-blocking verification devices that make you copy a set of fuzzy letters for security before registering on a site, are getting more and more annoying. Thanks to the increasing intelligence of character recognition technology, we're being expected to strain our eyes to make out whether something is a Q or an O, a J or an I, upper or lower case, and so forth. What, for example, is this? Words? I don't think so.
There's been a lot of talk in the past few months of alternative ideas involving object recognition... the idea being that humans do this much better than computers. Here are five promising approaches.
(...but as described this idea vulnerable to straight guesses. Spammers will take 1 in 9 attempts as success.)
(...addresses problem in the above, but - as executed here - some rotations hard to match even for humans.)
(Need to see wider variety of pics to judge.)
(Strikes me as both fun and exceptionally hard for a spam-bot to break... except by guessing right 1 time in 40,000.)
(This predates the others and to a layman seems ingenious. Michael tells me his patent application for his method is not yet through, and he has another even better idea about to be unveiled.)
Some more links:
P.S That often infuriating ReCapatcha technology illustrated at the top at least has a silver lining. You are
helping to digitize books