Machine Learning (Character Recognition)

Replies

• 

  rohit330

  Hi,

  There are python tools like Tesseract which are used for OCR however most of the time results are reliable when you work with high res images. When it comes to low res and if those are sort of predictable types one approach is to build a library of images of characters that you need to read & use exact comparison or root mean square difference. Alternately you could also scale up the image using various standard algorithm & then use Tesseract. That said, most of the times - working with different types of images will help you observe various challenges and accordingly decide on the right solution.

  Regards,

  Rohit

  
  

  
  

• 

  durga ch

  Neural networks?

  #-Link-Snipped-#

• 

  Anoop Kumar

  You need to look into how neural network work.

   Basically you need to train your modal (say your code) to recognize a given image to a character. You give a image to model which be taken as particular series of pixel as a character. The same you a human brain get trained.  More you train with different data, more accurate result you will get from test image.

  There is a Machine learning 101, course available on Coursera by top professor of Standford University. It's good starting for ML enthusiast.  

  For quick glance: #-Link-Snipped-#