iRis

OverView:

Today, visually impaired people’s access to most information depends on Braille books which are expensive and bulky. Hence, as a practical solution to this issue, we attempt to develop a text reader that can possibly alleviate the disabilities that deter the blind from accessing information. Presenting iRis - a novel method which can sequentially scan the text of any printed material, especially, non-digital material like paper, while providing real-time auditory feedback .The output can range from simple characters to whole words.
The prime motivation for this design is attributed to the preliminary studies conducted with visually impaired people. The prototype will be small scale and allow manageable operation with little setup.

Project Flow:

1.The program begins with detecting the fingertip wherein the user is asked to hold the finger steady for a few seconds until the tip coordinates are acquired. These coordinates are later used for finding the region of interest (ROI) for text extraction. A close-up view of the printed text is required as input for text extraction. We scan the frame continuously for words and detect all the words in a given frame. Thereafter we look for words that lie in the ROI and extract them. These words are then sent to the OCR Engine. The recognized text undergoes a dictionary check in order to ensure validity. Repetitive detection is avoided. If the recognized word is found valid, we invoke the TTS engine to utter the word.

2.Word recognition: The extracted word is resized and passed onto the Tesseract OCR for the recognition process (image-to-text conversion). The returned string undergoes a dictionary check. If a match is found, then the word is stacked up for text-to-speech conversion.

3.Text-to-Speech conversion: When new data is found in the stack, it is emptied and the text is passed on to the TTS engine for text-to-speech conversion and the word is uttered aloud. For transfer of data extracted to engine and between the two engines, we use a Last-In-First-Out (LIFO) approach by which we give higher priority to the most recently obtained data. If the program finds out that the user has moved ahead quickly and the earlier elements in the stack are of no use, then it is cleared and the process continues.

Unique and innovative attributes of your Project.

a. A wearable device a finger-mountable device, iRis, that enables reading of printed text. Using an intelligent
OCR-based algorithm coupled with real-time auditory and tactile feedback, iRis provides the user with an assistive text-reading experience. iRis is designed in such a way so as to permit easy setup-and-use with minimum calibration and high mobility.
b. Integration of Yocto - Edison.

What is the target audience and available market for your Project?
1. Visually impaired individuals and those with high difficulty in reading such as individuals with dyslexia.
2. Helps in educating individuals in various languages with the option to translate to various languages using the google engine.

How does your Project take advantage of Cloud connectivity?

As we have wi-fi module in Edison this is going to play the key role to transfer the data online to a web based server and to Cloud. IoT allows our product to connect to internet and translate between various languages.

Software components

1.Edison - Yocto
2.Node.Js
3.OpenCV
4.Tesseract OCR
5.Pyttsx
6.GTTS
7.Socket Communication

Hardware components?
1.Web Camera Module.
2. Buzzer
3.. LED Light

The Dev Kit?
Intel Edison Board.

Team -

Dishit Dak: ddak7579@gmail.com
Aswin Suresh Krishnan: aswinsureshk@gmail.com
Abel Saju Augustine: asa.11ec05@gmail.com
Hari Krishnan: hari.krishnan17@outlook.com