Let's talk captcha breaking (GOCR,TESSERACT,etc)

[WeissHut]

I have created my engine to break Google's internal captcha (Not recaptcha). It's accuracy is more 70%.

I have used, Tesseract Logic + Artificial intelligence + Neural network + 1 GB Captcha image database.

kalrudra, do you provide the source code, or a binary for solving google's internal captchas (not recaptcha)?

Thanks
WeissHut

 

[WeissHut]

Originally Posted by lwelch45
Using tesseract I coded an ocr server that processes images and returns the text from the image. Its currently in its beta stages and I will post back here with the sucess rate.


Hi lwelch45, you're not going on Google Internal Captchas (not Recaptcha), right?
Thanks
WeissHut

 

[falkor]

If any of you guys have managed to create a server based captcha decoder that could be used to solve peoples captchas online or via API let me know.

I'm currently looking to put a few hundred servers on the net that will solve captchas on a pay per solve basis.

I've got the technical expertise and startup cash but missing the key ingrediants to make the whole thing work

Feel free to drop me a pm or reply to this thread.

 

[-FPC-]

If any of you guys have managed to create a server based captcha decoder that could be used to solve peoples captchas online or via API let me know.

I'm currently looking to put a few hundred servers on the net that will solve captchas on a pay per solve basis.

I've got the technical expertise and startup cash but missing the key ingrediants to make the whole thing work

Feel free to drop me a pm or reply to this thread.

Just out of curiosity, why would you want to do that with Captcha Sniper out? Once it gets recaptcha working.. how will any one survive? Especially a new service?

 

[falkor]

Just out of curiosity, why would you want to do that with Captcha Sniper out? Once it gets recaptcha working.. how will any one survive? Especially a new service?

First time I'd heard of Captcha Sniper, although it looks great it dosn't support Twitter or Facebook which are the two main reasons people use mass captcha decoding based on stats from a friend at deathbycaptcha

 

[cody41]

What i'd like to see is some of the AI that will be developed to parse out what type of captcha is trying to be solved, then using the correct algorhythm or cleaning methodology of image to be used.

Assume I've got a list of 1000 captchas, of all types of platforms...phpboard, expressionengine, vbulletin, gmail, yahoo, etc..

Now, I feed those captchas into a sql db in one thread, then have my artificial intelligence identify the type of captcha it is, then apply the best cleaning method and then solve it back to the webpage.

THAT is the code I'm looking for or gameplanning. The db is necessary to handle transactions in and out, and to also take advantage of multithreading.

 

[The Data Guy]

well the server i created a while back was capable of doing simple things. And one of the joys of using a neural network is it allows you for better customization. It would be very easy to train a neural network to classify captchas into groups(recaptcha, pligg, etc). I was working on a pligg captcha solver and got it working at about 13 percent.

 

[Dally180]

well the server i created a while back was capable of doing simple things. And one of the joys of using a neural network is it allows you for better customization. It would be very easy to train a neural network to classify captchas into groups(recaptcha, pligg, etc). I was working on a pligg captcha solver and got it working at about 13 percent.

What are you guys coding these captcha solvers in? Are you using the .NE framework?

 

[The Data Guy]

I generally do most of my coding in vb.net but on occasionally i code in c++

 

[blakamia]

Just to bump this thread with interesting info:
Well I finally cracked recaptcha 80-90% accuracy. Used C# with Aforge:
http://code.google.com/p/aforge/

... for image processing and AI. No need to use stuff like Tesseract and GOCR, just code the individual character recognition yourself. It's very doable for current Recaptcha variant, but we will see when they update if I am able to keep up (Google always updates recaptcha when they detect someone cracked it, or they just generally shift it around to fuck with us).

 

[farlad]

blakamia, how much would I have to pay you to buy that recaptcha breaker off of you?

 

[truepals_forever]

I've a Recaptcha OCR with 40% accuracy

 

[farlad]

what's your price?

 

[premier]

I'm interested in your solving product can you PM me?

 

[tchalango]

It is not more easy to detect the letter from.the sound version of recaptcha?

 

[moonlighsunligh]

I think so , but you need something like Dragon naturally speaking or better for that, though if captcha is said letter by letter that is probably very simple.

 

[datomcat]

I've been using the free/open-source Julius speech recognition software to crack audio captcha's with some success. You can find it and pre-trained models on Sourceforge.

 

[tchalango]

Like in the picture version you have to separate the noise sound from.the speach frequency. I think because of the nature of sound separate different frequency is very easy. More than with picture but maybe i am wrong.

 

[farlad]

I've been using the free/open-source Julius speech recognition software to crack audio captcha's with some success. You can find it and pre-trained models on Sourceforge.

Are you preprocessing at all before using julius? Also, what commands are you using with julius? Thanks

 

[inamon]

Hi,

If you need a team of Captcha solvers consisting of as many as 20,000 workers, please PM me.
I would like to integrate a captcha breaking section on our CPA network where members can manually solve captchas for incentives. To capitalize this your Captcha serving site needs to have postback system.