Introduction

Image recognition now is getting very advanced with the deep learning techniques. It has a wide range of applications, e.g., autonomous cars and drones. The problem about the recognition devices for text applications is it confuses other languages with the targeted one and give the wrong prediction in some situations for instance translation or understanding street sign. Therefore, the project is focusing on making the recognition devices to be able to distinguish different languages so that it reduces the confusion errors of been different interpretations.

The motivation

Giving the suggested capability to recognition-devices will contribute more to improving image recognition devices that target text-related applications. Furthermore, the desire to work and practice on the image recognition with Raspberry Pi. Raspberry Pi is easy to use, inexpensive and set up. RPi reduces the complexity of the prototype. For most of the case if your project worked on RPi it will work on other alternatives.

The project workflow

Using Raspberry Pi camera to capture images of the font and OpenCV for real-time computer vision. The project will focus on five languages which are English, Japanese, Korean, Arabic and Hebrew. Knowing that there are different languages that look exactly like English which may require to implement test comprehension. These languages were selected for being different to simplify the complexity. To do the project first, we need to collect collections of images for each chosen language. Then train Raspberry Pi to recognize the font with the correct language.