The goal of this repo is to create implement different neural network architecture capable of recognizing text in text-line images containing lots of noises and background.
The first NN will be a CRNN architecture and it will be trained on generated images using TRDG.