How to use Bounding Boxes with OpenCV (OCR in Python Tutorials 03.02)

Поделиться
HTML-код
  • Опубликовано: 31 янв 2025

Комментарии • 40

  • @haniihsanuddin9585
    @haniihsanuddin9585 Год назад +13

    1. Blur image (to identify overall structure, and not focusing on text itself)
    2. Create threshold (and kernal) to separate text block
    3. Perform dilation (~white thickening)
    4. Perform contour (finding boundaries)
    5. Perform loop to only draw boundrary box of specific size (to exclude small bbox)

  • @letslearn2674
    @letslearn2674 2 года назад +3

    This is the one I have been looking for. Thank you so much!

  • @aayushsinha7439
    @aayushsinha7439 7 месяцев назад

    Thanks for such a simplified explanation, helped me with my ongoing project a lot!

  • @BrandonJF4
    @BrandonJF4 2 года назад +2

    Thank you so much, this really helped me make progress on a project!

  • @fuemma--7122
    @fuemma--7122 4 месяца назад

    The opencv was so easy to understand!

  • @thenotoriousrkf3012
    @thenotoriousrkf3012 3 года назад +16

    I guess, there is an error in your code. From minute 15:45 on, you define the ROI. However, instead of x+h, w would have to be added to x. Therefore, roi should be defined as: roi = image[y:y+h, x:x+w]
    Since this typo also appears on your GitHub you should change it there as well.
    Kind regards!

  • @niladrimallik3172
    @niladrimallik3172 4 месяца назад

    At 15:29, after adding the "if h > 200 and w > 20:" statement, I am still getting the same result as without the if statement. Any idea why this is happening? I changed variable names, defined the contours again, but still the same result.

    • @Maruti_Pai
      @Maruti_Pai 4 месяца назад

      rerun the whole code again

  • @Atharva_S9
    @Atharva_S9 3 года назад +8

    why can't you provide the code for this

  • @kltr007
    @kltr007 Год назад +3

    Short question: in Box [15] it reads "else cents[1]". Is this a typo and should be "else cnts[1]" or did I miss something?
    But great content! Keep going!

    • @pcb5135
      @pcb5135 Год назад

      i assume its typo

  • @joshuasmitherman1712
    @joshuasmitherman1712 Год назад +3

    It's not finding the sections for me. It captures the whole document as a section. Any suggestions?

    • @ThatRussian
      @ThatRussian 5 месяцев назад

      Did you find the solution?

    • @ThatRussian
      @ThatRussian 5 месяцев назад

      I actually found the solution if you're still interested: you basically need to crop the image so no extra blank spaces are left. Since mine was a vertical page I used this code - image[10:1060, 670:1250] image[start_row:end_row, start_column:end_column]

  • @ppmanguin
    @ppmanguin 2 года назад +1

    at 11:16 I have an error, can you tell me how to fix it? Thank you!
    error: OpenCV(4.7.0) :-1: error: (-5:Bad argument) in function 'boundingRect'
    > Overload resolution failed:
    > - array is not a numerical tuple
    > - Expected Ptr for argument 'array'

    • @ppmanguin
      @ppmanguin 2 года назад

      fixed by adding a variable, because findContours creates 2 outputs.
      cnts, new_variable = cv2.findContours(dilate, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)

  • @steffenhalama5558
    @steffenhalama5558 3 года назад +1

    Very nice video helped me a lot.

  • @conorforster8853
    @conorforster8853 2 года назад +3

    Hi, this tutorial series has been the best thing slince sliced bread, and honestly dont know where id be with out it
    however i am stumped, im trying to read pdfs into jpeg format, the problem arises when i have tables and images within these files that i would like to either skip or try to read into file with out wreacking structure (obviouslty not images within the images). idealy i would like this process to be automated as the final program is not being used by myself but by others less aquainted with technolagy. As of now there is no documentation i can find that helps facilitate this.
    i know its a long shot but honestly ive hit a wall and if by some chance anyone can help and guidence would or advice would mean the world

  • @vildanhuseynov6492
    @vildanhuseynov6492 3 года назад +1

    good job man!!!

  • @ridafatima1739
    @ridafatima1739 9 месяцев назад

    will this work on colored images as well , if not, what changes should I make for the colored images?

  • @virendartripathi4645
    @virendartripathi4645 Год назад

    I can't download the images from the course can you help me so that I can practice this

  • @khushibaghel220
    @khushibaghel220 Год назад

    I am trying to run this in google colab but getting an error: TesseractNotFoundError: C:\Program Files\Tesseract-OCR is not installed or it's not in your PATH. See README file for more information. How to resolve this? I have already added pytesseract in my env variables

  • @mateussaar4071
    @mateussaar4071 11 месяцев назад +1

    MAGIC

  • @DilipDas-ys5ph
    @DilipDas-ys5ph 2 года назад

    Great Thanks !!

  • @uswakhan3050
    @uswakhan3050 3 месяца назад

    how to apply ocr on different language text

  • @breezyfeels1802
    @breezyfeels1802 2 года назад

    Hi, can you please tell how I can have bounding boxes around each question in any question paper? I have tried a lot, but unable to get it. I would be really glad if you could help me..Thanks!

  • @jumbertparrenas3218
    @jumbertparrenas3218 2 года назад

    Can I ask this is capable to application or only for desktop..? Im asking because this is same on my title thesis.

    • @nathantafelsky7089
      @nathantafelsky7089 Год назад

      It could be used in the source code of an application, or used on different operating systems. Imports and syntax would vary by language and implementation.

  • @farahjabeen7707
    @farahjabeen7707 3 года назад

    @Python tutorials for digital humanities can you explain how to make bounding box using pixel location?

  • @anjuathouse5370
    @anjuathouse5370 3 года назад +1

    could you please make a video on handwritten scanned document image line segmentation

    • @python-programming
      @python-programming  3 года назад +2

      Sure! I actually wrote that code a year or so ago. I will try and dig it up and make a video on it.

    • @anjuathouse5370
      @anjuathouse5370 3 года назад

      @@python-programming Thank you so much..

  • @vildanhuseynov6492
    @vildanhuseynov6492 3 года назад

    dude, do you have experience in aligned text?

  • @rapidash1995
    @rapidash1995 18 дней назад

    ❤❤❤❤

  • @ROKKor-hs8tg
    @ROKKor-hs8tg Год назад

    كيف يمكن عرض اشكال مطبوعة ف صورة ممسوحة ضوئيا الى ملفdocx

  • @tiennguyentran9358
    @tiennguyentran9358 3 года назад

    *i Love u so much tks u*

  • @darksidegumball7205
    @darksidegumball7205 6 месяцев назад

    I am sorry, but you keep saying that i explained these things on the previous videos, and i watched all of the previous ones and all you did was copy code and paste it into your juypter notebook without any proper explanation, hope you can provide a newer tutorial, otherwise thanks for the tutorials.