Re:Unredacter: Never ever use pixelation as a redaction tech...
by danbruc on Saturday December 17, 2022 @03:53PM
(#34032129)
This also seems like something a [convolutional] neural net should be able to learn pretty well and generating training data could be fully automated. Throw everything at it - different fonts, different sizes, bold, italic, different colors, different anti-aliasing methods, sub pixel offsets, pixelation sizes and offsets, JPEG noise, rotations, ... - and maybe also give it some language model to better reason about plausible letter combinations. I would love to see, how good this could be.
[ Parent
| on HN ]
Re:Unredacter: Never ever use pixelation as a redaction tech...
by buildbot on Saturday December 17, 2022 @04:32PM
(#34032463)
Already been done! https://github.com/HypoX64/DeepMosaics
Re:Unredacter: Never ever use pixelation as a redaction tech...
by foota on Saturday December 17, 2022 @07:09PM
(#34033655)
Image models are pretty bad at text, generally, so I question how well it would work on text.
Re:Unredacter: Never ever use pixelation as a redaction tech...
by buildbot on Saturday December 17, 2022 @10:00PM
(#34034673)
Great point!
Re:Unredacter: Never ever use pixelation as a redaction tech...
by mft_ on Sunday December 18, 2022 @05:51AM
(#34036521)
It’s not, though. It seems to be a very shallow understanding of the subject (or just insufficient consideration?) stated confidently (as so often happens these days).
As a sibling comment to yours also points out, there are plenty of examples to the contrary - models taught to recognise text from indistinct images. We’ve all been contributing to this for years via Captchas, and now we’ve got to the stage where our iPhones apply the concept locally to text found in our photo libraries.
Re:Unredacter: Never ever use pixelation as a redaction tech...
by buildbot on Monday December 19, 2022 @01:31PM
(#34055699)
It was my understanding that anything trying to recover text is still hard to do well, but I could be wrong - I have not worked with image models in awhile. Classification and OCR is truly well solved though.
Re:Unredacter: Never ever use pixelation as a redaction tech...
by teruakohatu on Sunday December 18, 2022 @04:03AM
(#34036117)
You are probably thinking of diffusion generation models. Models in general work absolutely fine with text. Just about the first convolution neural network examples you come across are basic character classification examples.
Re:Unredacter: Never ever use pixelation as a redaction tech...
by layer8 on Saturday December 17, 2022 @06:16PM
(#34033298)
For blurred (instead of pixelated) text, take a look at https://arxiv.org/abs/2211.10103.
Re:Unredacter: Never ever use pixelation as a redaction tech...
by PartiallyTyped on Sunday December 18, 2022 @04:40AM
(#34036255)
Should be very straight forward with cold diffusion.
Re:Unredacter: Never ever use pixelation as a redaction tech...
by loh on Sunday December 18, 2022 @05:59AM
(#34036554)
A neural net was my first guess as to how this would work but was pleasantly surprised at the actual technique used, which makes a lot of sense. The animated gif does a great job of explaining it on its own.