What you will learn
- Use Tesseract, EAST and EasyOCR tools for text recognition in images and videos
- Understand the differences between OCR in controlled and natural environments
- Apply image pre-processing techniques to improve image quality, such as: thresholding, inversion, resizing, morphological operations and noise reduction
- Use EAST architecture and EasyOCR library for better performance in natural scenes
- Train an OCR from scratch using Deep Learning and Convolutional Neural Networks
- Application of natural language processing techniques in the texts extracted by OCR (word cloud and named entity recognition)
- License plate reading
Requirements
- Programming logic
- Basic Python programming
Description
Within the area of Computer Vision is the sub-area of Optical Character Recognition (OCR), which aims to transform images into texts. OCR can be described as converting images containing typed, handwritten or printed text into characters that a machine can understand. It is possible to convert scanned or photographed documents into texts that can be edited in any tool, such as the Microsoft Word. A common application is automatic form reading, in which you can send a photo of your credit card or your driver’s license, and the system can read all your data without the need to type them manually. A self-driving car can use OCR to read traffic signs and a parking lot can guarantee access by reading the license plate of the cars!
To take you to this area, in this course you will learn in practice how to use OCR libraries to recognize text in images and videos, all the code implemented step by step using the Python programming language! We are going to use Google Colab, so you do not have to worry about installing libraries on your machine, as everything will be developed online using Google’s GPUs! You will also learn how to build your own OCR from scratch using Deep Learning and Convolutional Neural Networks! Below you can check the main topics of the course:
- Recognition of texts in images and videos using Tesseract, EasyOCR and EAST
- Search for specific terms in images using regular expressions
- Techniques for improving image quality, such as: thresholding, color inversion, grayscale, resizing, noise removal, morphological operations and perspective transformation
- EAST architecture and EasyOCR library for better performance in natural scenes
- Training an OCR from scratch using TensorFlow and modern Deep Learning techniques, such as Convolutional Neural Networks
- Application of natural language processing techniques in the texts extracted by OCR (word cloud and named entity recognition)
- License plate reading
These are just some of the main topics! By the end of the course, you will know everything you need to create your own text recognition projects using OCR!
Who this course is for
- Anyone interested in OCR (Optical Character Recognition)
- Undergraduate students who are studying subjects related to Artificial Intelligence, Digital Image Processing or Computer Vision
- Data Scientists who want to increase their knowledge in Computer Vision
- Professionals interested in developing professional optical character recognition solutions
- People interested in creating their own custom OCR
The instructor goes quite slow on the explanation. It was hard to keep up on all the lecture. The lessons could use more energy and interesting examples, to make it more fun and exciting. The accent was heavy, sometimes harder to understand his english.