0

I'm working on a startup, and I believe I've found two infringing patents about removing text from shareable content using OCR and machine learning. These patents are US2020012817 & US2020042837 and my prior art is at https://benjaminhowarth.com/archive/2019/03/11/introducing-trash-panda-a-tool-for-masking-text-from-screenshots and at https://github.com/benjaminhowarth1/Trash-PaNDA

  • Just want to point out that the linked documents are applications, not patents. You are allowed to make the patent authorities aware of your prior art. – Eric Shain Mar 12 at 16:56
  • Also, it is not clear what you mean by the word "infringing". – George White Mar 12 at 21:35
  • It’s important to focus on the claims to determine if your technique would infringe in these applications. Also, the claims will likely narrow before (and if) an actual patent is granted. – Eric Shain Mar 13 at 1:06
0

I see the date on your article as 11.03.2019. If the date is in the form month.date.year, it was after both applications.

The application that US2020012817 was filed August 8 2019, so before November 3rd 2019. The other application was filed in 2018, so your article was not prior to it either.

| improve this answer | |
  • 1
    George, the publishing date of his article is March 11, 2019 and not November 3, 2019. See the date format on the main page of his blog benjaminhowarth.com – Mikk Putk Mar 13 at 7:24
  • Thanks George - if you double-check the Github commit history, it disambiguates the date clearly, the initial code commits are on 11th March 2019. – Jamie Howarth Mar 13 at 16:18
  • @MikkPutk - thanks, that makes the date format clear and it is before one of the applications was filed. Now the question is how the disclosure relates to the claims. – George White Mar 13 at 16:56
0

It’s important to focus on the claims to determine if your technique would infringe in these applications. Looking at the first claim of the earlier submission:

  1. A tangible, non-transitory, machine-readable medium storing instructions that when executed by one or more processors effectuate operations comprising: receiving, with one or more processors, a screen capture event from an operating system of a first client computing device of a first user, the screen capture event including, or being associated with, a bitmap image of at least part of a display of the first computing device; causing, with one or more processors, optical character recognition (OCRing) of text in the bitmap image and obtaining, as a result of the OCRing, an OCR record with text appearing in the bitmap image and indicating locations of characters of the text in the image with coordinates of pixels in the bitmap image, the text comprising a plurality of n-grams; scoring, with one or more processors, each of the n-grams based on whether the respective n-grams match any of a plurality of patterns; classifying, with one or more processors, each of the n-grams based on the scoring into two or more categories, the two or more categories including a category for confidential information; for each of the n-grams classified in the category for confidential information, with one or more processors, obfuscating the respective n-gram in the bitmap image to form a modified version of the bitmap image with operations comprising: determining, based on the OCR record, coordinates of pixels in the bitmap image corresponding to the respective n-gram; and modifying, with one or more processors, pixel values in the bitmap image of pixels at the determined coordinates in response to determining the coordinates of pixels in the bitmap image corresponding to the respective n-gram; storing, with one or more processors, the modified version of the bitmap image at a remote server system; providing, with one or more processors, from the remote server system, to the first user computing device, a uniform resource identifier at which the modified bitmap image is accessible; receiving, at the remote server system, from a second user computing device, a request for the bitmap image at the URI; and sending, from the remote server system, to the second user computing device, the modified version of the bitmap image.

I'm assuming "URI" really should be "URL". In any case, the claim's novelty seems to hang on the term "n-grams". This isn't my field and I didn't read the rest of the application so I don't know what an "n-gram" is beyond what is described in Wikipedia. Based on that article it would seem pretty obvious to apply them to this task. To infringe on a patent you need to implement each and every step in at least one claim.

So, one possibility is that your technique doesn't employ n-grams as described by the application. Another is that the use of n-grams is either well know in similar applications prior to the submission date or else the use of n-grams to solve this problem is obvious to someone skilled in this field. You are in a better position to assess this than I am. Based on my very minimal search, it seems that the use of n-grams in OCR is well known and probably has a lot of prior art. For instance a quick search on Google Scholar provided this paper from 2002 and this paper from 1997. If so, I'm not sure that just blacking out the results of OCR constitutes a novel idea. Also, I see from your Github link that you are utilizing Microsoft's Computer Vision API. If that code predates the application's filing date it also constitutes prior art.

Please note that the claims will likely narrow before (and if) an actual patent is granted.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.