harishdeepak / image-based-character-recognition-documentation-system-to-decode-inscription-from-temple

trainining code for custom data in tesseract

data :

    add pre-processed images(jpeg,jpg,tiff,png) here

scripts :

    purge.py  -- use it to clear previously trained files for a fresh start/n
    
    rename.py -- the input images needs to be renames in the format <language>.<font>.exp<i>.png
    
    boxgen.py -- generate box files for our images.
   
    openjTess.py -- open jtessbox editor to edit miss predicted box files
   
    trainimgs.py -- run this train all the images in a cnn network

testout:

    output of test.py gets stored here

trainfiles:

    contains important train files and properties files necessary for training

trainoutput:

    contains the .traindata file and other intermediary output file.