1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Search For Text Within Images

Discussion in 'Black Hat SEO' started by Moonlight1, May 29, 2017.

  1. Moonlight1

    Moonlight1 Newbie

    Joined:
    May 14, 2017
    Messages:
    20
    Likes Received:
    2
    Gender:
    Male
    Is there a tool or way to search for specific text within a folder full of images in Windows?

    I've scraped content from some IG accounts and I don't want to use the images that have branding.
    Not all images have it and I'd rather not search in 9000 images for a couple of branded text within images.

    Any way to do this?
     
  2. Grimmmm

    Grimmmm Newbie

    Joined:
    May 4, 2017
    Messages:
    13
    Likes Received:
    2
    Gender:
    Male
    Well idk if there Exists that kind of tool... but you can write by yourself or buy a script from someone else, that is capable of OCR,

    P.s the script can be easily made by Using pytesseract...
     
  3. Mojo Jojo

    Mojo Jojo Newbie

    Joined:
    May 24, 2017
    Messages:
    17
    Likes Received:
    0
    If you aren't familiar with machine learning then you'll have to google around but I highly doubt that you will find a pretrained network that suits your needs.

    If you have experience with ANN's (Artificial Neural Networks) then here is what I suggest.
    Pick 100 images with text per hand, use it as sample data. The rest as training and validation data. Model of choice would be RNN (recurrent neural network) or maybe even something fancier like the BDLSTM (Bi Directional Long short term memory).

    If all of the images with text share a similiar structure, like the same font or the same place where the text is placed, then you won't even have to fool around a lot to find the right topology and parameters for your network.

    Good luck!
     
  4. Moonlight1

    Moonlight1 Newbie

    Joined:
    May 14, 2017
    Messages:
    20
    Likes Received:
    2
    Gender:
    Male
    Thank you for all the input but creating something myself is too complex.

    Is there is not a tool that scans your folder with ocr and makes it able to search within these images on a specific text? Automatically is not necessary, some manual searches is fine.