The model predicts handwritten texts to actual texts. The input to be provided is a picture of an image containing a word and output is the text.
We can segment the words in the image:
Crop the words into different imaages and feed to the model.
Output in /src/words.txt