Words are a powerful way to communicate with others. However, some people do not have the ability to read the words on a document or book. With the advance of technology, people have found ways to solve problems that challenge humans on a daily basis. We wanted to give a solution to those who are visually impaired. We now provide a way to interact with a computer that reads the words from a detected image.
What it does
The purpose of this program is to output the words it detects from the webcam. This will provide a way for those with visual disabilities to be able to know what words are on a document, page and book. Our purpose is to help those with visual disabilities to be able to experience words in a whole new way.
How we built it
We built the application in Python. We use OpenCV-python to implement the computer vision aspect of this program. Open-CV handles object tracking. We use pytesseract-ocr to recognize the text in the images that the camera captures. We use Pillow (an updated version of the now deprecated PIL) as an image processor. We use the built-in python text-to-speech tools.
Challenges we ran into
We have a working prototype that works with only large text, black and white text. It cannot read smaller fonts, multiple lines at the same time, and cannot handle large, frequent, hand movements.
Accomplishments that we're proud of
We're proud that it works! It is currently a work-in-progress, however, we're proud that it can actually recognize text, input it into the system, and output with python text-to-speech.
What we learned
Our team had both new programmers and seasoned programmers. New programmers learned how to use tools such as Git (Version Control), programming in python, installing packages, and using a command line. Our seasoned programmers learned more about python tools such as OpenCV, OCR, and other tools that are used for computer vision.
What's next for Cicero
We would like to improve on how the camera will read text and handwriting.