Optical Character Recognition - Machine Learning
OCR (Computer Vision)
OCR is a project in the image processing field. The project is based on an algorithm, which allows to analyze a given image and to find out where text is placed on it.
The algorithm, at beginning, analyzes the picture using Convolutional Neural Networks (CNN), provides results (an amount of features) to Recurrent Neural Network (RNN),
which predicts character on each feature and, at least, Connectionist Temporal Classification (CTC) converts to the text label. (Also, it is possible to use other types of classification algorithms such as Naїve Bayes, Hidden Markov Model, etc.)
Usage of OCR algorithm: conversion of handwritten text to machine-readable; extracting contact information; passport recognition in airports; aid for blind people; as preprocessing step in more complex systems such as license plate recognizer.