AutoHotkey V2- OCR (Optical character recognition) Image to Text

Поділитися
Вставка
  • Опубліковано 14 гру 2024

КОМЕНТАРІ • 16

  • @TABNationAutomation
    @TABNationAutomation  2 місяці тому +3

    What else should we look at when it comes to OCR?

    • @steelseeker
      @steelseeker 2 місяці тому +2

      do findtext v2 i find it alot better for ocr

  • @dunnerca
    @dunnerca 2 місяці тому +1

    The screen-snip example is great. I'd like to give the option of getting rid of line breaks so the text can be imported as a paragraph of text. I can easily see how these scripts can be really useful day-to-day.

    • @TABNationAutomation
      @TABNationAutomation  2 місяці тому +1

      @@dunnerca you could look at string replace to get rid of breaks

    • @dunnerca
      @dunnerca 2 місяці тому

      @@TABNationAutomation Thanks Thom. I followed your tip and it worked great. I actually changed it so the line feeds in the clipboard contents are stripped out but it keeps the separate paragraphs. I had to do it in three steps. First, I did StrReplace to look for two line feeds in a row and replaced them with a new paragraph variable ("XXXNP"). Then I did a StrReplace to strip out all the line feeds. Finally, I did a StrReplace to replace the "XXXNP" with a line feed. It may not be the most efficient way but it works. In my job (CPA), I have to convert a lot of PDF's to Excel tables. Normally I do that in ChatGPT. My next task will be to modify this script to copy the data to an Excel sheet but put the data in separate columns. That will be much easier than going through ChatGPT first. Many thanks.

  • @thepragmatic6383
    @thepragmatic6383 2 місяці тому

    I tested this tool on your video, while you were showing portions of the script used.
    I paused your video at 3:36 and selected lines 30 and 31 of the script visible on the screen. Here is the result obtained:
    30 resu It := OCR. FromBitmap (hBitmap, , 2)
    31 : = rearrangeOCRresu1t (result)
    The copy contains some errors, but the result is still impressive.
    OCR seems to have difficulties with the letter "l or L", as well as with the purple color by omitting the word "text" from line 31.

    • @TABNationAutomation
      @TABNationAutomation  2 місяці тому

      Something I have done in the past is after the OCR, take the text and run it with some AHK spell check code. This helps to find mistakes.

  • @MrSawok
    @MrSawok 2 місяці тому +1

    Hey hey! Big thanks Tom ❤

  • @tinhdaika3931
    @tinhdaika3931 2 місяці тому

    thanks bro

  • @gabisoare941
    @gabisoare941 2 місяці тому

    Hello, can I modify several keys to transmit different texts? Ex: key [ to write the name, key ] to write the first name, key { to write the date of birth, key } to write passport number? They would be necessary for the quick filling of some online forms. Thank you.

    • @TABNationAutomation
      @TABNationAutomation  2 місяці тому

      this is called hotstrings. I did some videos on it but simply
      ::f1::my name

  • @vipulshivmore3764
    @vipulshivmore3764 Місяць тому

    I need to use the snipping function in my office but there are lot of restrictions with the IT dept. there, can i use the codes in a normal notepad and run this, how it will work. am a non tech guy, can someone help me on this.

    • @TABNationAutomation
      @TABNationAutomation  Місяць тому

      You need to install ahk. Depends on your company but I did a video about how to pitch ahk to your job

  • @dked-u9i
    @dked-u9i 14 днів тому

    Is Korean supported?

    • @TABNationAutomation
      @TABNationAutomation  14 днів тому

      I have no clue what your computer can do. If you read the documentation it tells you how to check