INTRODUCTION

In recent years, we have heard the media mention Artificial Intelligence and we are witnessing a period of its explosion, with pioneering companies in this field such as Google, Facebook.

Artificial Intelligence has developed to the point of completely replacing humans in certain jobs, even better with a large enough amount of data it uses.
However, artificial intelligence is not new, it has been researched since the 50s of the last century with a typical application being chess playing machines. In the period from 80s to 10s, machine learning began to be interested and used a lot, typically search engines and spam filters.

After 2010, with the rapid evolution of hardware architecture, which made the cost of computing decrease rapidly, deep learning technology, or multi-layer neural networks, was created to explode, one of the remarkable capabilities of this technology is the ability to recognize images. By 2015, the accuracy had surpassed human capabilities. Another advantage of this technology is that it is very user-friendly. The neural network can be considered a black box, with the task of recognizing images, just show a sample image, the network automatically encodes similar features into parameters, and uses those parameters to predict new images.

The Development of Artificial Intelligence

Based on this new technology, we have successfully developed a letter and number recognition system. AVA's OCR software can learn different types of characters and perform handwriting recognition based on what has been learned. The OCR-AVA software can be easily integrated into customers' handwriting recognition systems. We have developed a user-oriented interface for flexible and convenient operation when customers need to read a new font.

Basic steps in the process of building a text reading application of ours:

AVA-OCR software has been tested and has high quality, the rate of reading numbers on phone cards correctly is 95%.