3 ms·
This a great writeup and super easy to follow along with. The figures are really nice! One observation: training a neural net to classify segmented characters
by lstyls 12y ago
This a great writeup and super easy to follow along with. The figures are really nice!
One observation: training a neural net to classify segmented characters is probably overkill. The author observed that the font never changed, but never ended up exploiting this fact. After the very effective preprocessing, thresholding, etc the characters are almost identical to the 'average' representations the author generated!
I bet it would be enough simply to classify an unknown character by the letter that it shows highest correlation with.
- malgorithms 12y agoAgreed! Also, given the successful character extraction, and the knowledge that (1) the font doesn't change, and (2) they're just translated and rotated, I think performing those operations on the individual characters could've yielded a pretty perfect success. Simply try a whole bunch of shifts and rotations on a given character until it matches a reference, almost exactly.
- mieko 12y agoThis is all true. The first thing I built was an awesome hammer, and I had a large amount of nails to deal with. In retrospect, I'm sure there was a good deal of more focused, surgical solutions that'd save some overhead.
- throwaway15213 12y agoHe added the corpus to the repo so you can try it: http://www.reddit.com/r/programming/comments/2hisfk/breaking_the_silk_roads_captcha/ckt60k9 http://www.reddit.com/r/programming/comments/2hisfk/breaking...
- contingencies 12y agoSame thinking here: IIRC there are some pretty good open source OCR programs that would make short work of the individual characters, with no need to re-assemble and/or neural networkify. However, the boilerplate code he gives is generally applicable to less braindead captures, which is a great resource for others.