Lets discuss ScrapeBOARD. Questions, comments, thoughts, Videos etc..
- Thread starter loopline
- Start date
-
- Tags
- scrapeboard scrapebox
- Status
would be wonderful if this had the decaptcha like xrumer integrated..how can it be that just these guys have automatic decaptcha?
danhoff
Senior Member
- May 11, 2010
- 802
- 62
question:
When i create accounts i get for ex.:
3 000 accounts activaed
6 000 email activated accounts
70 moderated
When scrapeboard downloads activation emails and click on the links, it sayd i have 6 000 activated accounts. The file with accounts also has 6 000 accounts.
Shouldn't it be 3 000 already activated accounts + 6 000 from email, so around 9 000 acounts in total in the file ??
When i create accounts i get for ex.:
3 000 accounts activaed
6 000 email activated accounts
70 moderated
When scrapeboard downloads activation emails and click on the links, it sayd i have 6 000 activated accounts. The file with accounts also has 6 000 accounts.
Shouldn't it be 3 000 already activated accounts + 6 000 from email, so around 9 000 acounts in total in the file ??
Last edited:
kokoloko75
Elite Member
- Jan 1, 2011
- 1,627
- 1,950
UPDATE : Ok, forget... Without proxies (however fresh) it's works.
For me, ScrapeBox and ScrapeBoard URL hervester dosen't works today...
Where can this problem arise ?
Someone in the same case ?
Beny
For me, ScrapeBox and ScrapeBoard URL hervester dosen't works today...
Where can this problem arise ?
Someone in the same case ?
Beny
Last edited:
Just bought this earlier this evening.
I've scraped some keywords and then used that to scrape over 10K urls. After trimming out duplicates I end up with 250.
I then ran these through the Forum analyzer and when I try to export the results filtering out 'non public profiles' the text file I export to is blank.
Am I doing something wrong?
I've scraped some keywords and then used that to scrape over 10K urls. After trimming out duplicates I end up with 250.
I then ran these through the Forum analyzer and when I try to export the results filtering out 'non public profiles' the text file I export to is blank.
Am I doing something wrong?
I for one love the idea of an integrated Captcha breaker. The problem with Scrapeboard is that it's not possible to train the breaker in an easy way.
I've been looking into how I would combine the AntiReCaptcha (arc.traineddata) code with the current eng.traineddata and just can't find a good answer. I'd love it if:
1) It could be integrated into SBoard, if it isn't already (jDownloader has this library and trained data)
2) The possibility to train with the Tesseract tools available. This could come in handy both for one self and to eventually crowd-source the data, thus building a really neat captcha breaker.
Then comes the request for the text-captcha breaker. Not everything is possible to do here (example: what is this sites name?) from a generic answer, and that's not what I'm really looking for either. But it would be great to be able to edit, add and delete answers. This feature, coupled with the crowd-sourcing already in place, would make Scrapeboard the number one champion.
If anyone knows how to merge, train and fix the tesseract trained data for several captcha types - please reply!
I've been looking into how I would combine the AntiReCaptcha (arc.traineddata) code with the current eng.traineddata and just can't find a good answer. I'd love it if:
1) It could be integrated into SBoard, if it isn't already (jDownloader has this library and trained data)
2) The possibility to train with the Tesseract tools available. This could come in handy both for one self and to eventually crowd-source the data, thus building a really neat captcha breaker.
Then comes the request for the text-captcha breaker. Not everything is possible to do here (example: what is this sites name?) from a generic answer, and that's not what I'm really looking for either. But it would be great to be able to edit, add and delete answers. This feature, coupled with the crowd-sourcing already in place, would make Scrapeboard the number one champion.
If anyone knows how to merge, train and fix the tesseract trained data for several captcha types - please reply!
Thanks for the answer. I'm not expecting miracles in the near future, I see SBoard as a fun tool that's actually useful. Have totally massive scraped lists for forums that I just let it burn through.
Regarding the progression - I would actually say that I'd probably crowd-source as much as I can. The benefits are several.
The most obvious is of course that you can let others "work" for you. Let them train the captcha breakers while you make all the fun features that keeps you motivated.
The second reason is for the users themselves. Lots of people bought SBoard instead of Xrumer and several will have done so because of a tight budget. If you just give them a small carrot, they'll put work into captchas and get rewarded for doing so.
This will of course also raise a buzz - if it's locally trained then people will start sharing, buying etc and get even more users to SBoard.
If it's a public crowd-sourced training then the buzz will come from having a really good captcha breaker.
Of course, how does one protect the investment with an open database? I see gazillions of captcha-breaking services rising from a crowdsourced database.
Ah well, just my two crowns!
By the way, does anyone have good trained response files to share? Mine are... well, under average. I think that comes to play with SOME of the missed activation emails.
Just bought today, out of 2000, 1 got activated. I manually did the captcha etc,
Most common reason: unknown platform and errors, such as 404, etc. Other reasons were like non public profile (maybe less 30)
Matt: there really needs to be more info in the online help guide. On the video they were helpful dont get me wrong, but certain parts you seem to rush it, and when it is written it lot easier to refer to the part I need, instead having to fast forward, plus you can get more info down.
Most common reason: unknown platform and errors, such as 404, etc. Other reasons were like non public profile (maybe less 30)
Matt: there really needs to be more info in the online help guide. On the video they were helpful dont get me wrong, but certain parts you seem to rush it, and when it is written it lot easier to refer to the part I need, instead having to fast forward, plus you can get more info down.
Last edited:
- Jan 25, 2009
- 6,378
- 3,774
Well mate I don't write the online help. I just do videos. You are correct, I do tend to talk fast. Its not that I am trying to rush it, its just kind of how I am.
I am going to be redoing all the videos, probably a couple of times yet. Since its beta I wanted to get some videos done, but so much changes so fast I could do videos this week and next week do them again.
Right this minute I can't do any videos as I had surgery 2 weeks ago and then again just monday of this week. So when I talk for more then about 3 seconds you can hear the pain in my voice and that would make for bad videos. Plus videos take a fair bit of concentration and preparation and I just don't feel up to it at the moment.
Also bear in mind the developer is writing the help file. They are spending their time on making a great product right now, so they can officially launch the program at some point. So I am sure after they get a lot of that done they will be adding more info to the help file. Its all still a work in progress.
I will commit to slowing down though, in future videos anyway.
Thanks,
MAtt
pyronaut
Senior Member
- Dec 9, 2008
- 1,189
- 1,440
Cool to have someone who knows what they are talking about
Just from my personal understanding, and me using it, Tesseract has severe memory leaks. Whether they are so bad that it is obvious, or just slow leaks over time, I do not know. I have only ever used tesseract in a .net assembly, and I did not notice it that bad, but I was doing only small tests.
Next, you need to understand that tesseract is only ever used for reading PDF's or straight images. Not captchas. You will need to code the cleaning of the image and straighten out the letters etc before passing it to tesseract. You can't just throw a random captcha in there (Either recaptcha or any other) and expect it to read it (Because it won't).
A few more things. Tesseract was a tonne more accurate when splitting letters up in a captcha, and feeding the letters individually. Trying to split letters up in a Recaptcha sense can be difficult, and is not always 100%. Not a deal breaker, but your success rates increase 10 fold.
And lastly. Although this could have been in the implementation I was using. Tesseract seemed "Random". 1/20 solves on the exact same captcha would return a wrong result. If I ran it again, it would come back with the correct result. I have no idea what causes it, but it certainly isn't 100%.
softtouch2009
Senior Member
- Dec 2, 2009
- 1,064
- 257
ScrapeBoad does cleaning up images, straighten text and split the image into individual captchas before they are feed to tesseract. So, training tesseract might not help at all.
@Pyronaut: About your random results, do you save the image uncompressed or use jpeg? We do not get random results, every time we test the same image, the result is always the same. We use .jpg images, but without compression (quality 100%).
@Pyronaut: About your random results, do you save the image uncompressed or use jpeg? We do not get random results, every time we test the same image, the result is always the same. We use .jpg images, but without compression (quality 100%).
- Jan 25, 2009
- 6,378
- 3,774
Hey mate, I don't work for scrapebox/board, I simply maintain a "quality" relationship with the dev team. I make it a point to not go around saying anything i am told. If they want it public, they will share it. Not that I have a checklist anyway, but I am fully confident that they have the application well planned out and will do everything they can to make scrapeboard and excellent application.
pyronaut
Senior Member
- Dec 9, 2008
- 1,189
- 1,440
They were directly downloaded SMF captchas. Of which, my app then turned them into "bitmaps" in memory (Just the way it is in .net).
If Scrapeboard is already cleaning images, then you don't need to "train" it at all. The english language pack it comes with read SMF captchas just fine after cleaning it up a bit.
bearmanpig1337
Newbie
- Jun 20, 2011
- 15
- 1
keizer
Regular Member
- Oct 22, 2008
- 397
- 402
I am waiting to buy but when it is complete. Will you ask developers how long it could take to be full functional.
Well nobody answered my question in the thread I created so I'm quoting it here :
Hope you guys can enlighten me.
Hope you guys can enlighten me.
softtouch2009
Senior Member
- Dec 2, 2009
- 1,064
- 257
sbox used your proxies. But, before you connect to any proxy, you will pass through your ISP of course.
YOU -> YOUR ISP/GATEWAY -> PROXY -> TARGET SITE
So, your ISP see that the traffic is coming from you.
YOU -> YOUR ISP/GATEWAY -> PROXY -> TARGET SITE
So, your ISP see that the traffic is coming from you.
- Status