A data scientist in India created a computing model based on the NVIDIA GPU that is able to translate Braille to text and audio, Indiatimes reported. Bhavesh Bhatt used his prowess in data science, machine learning, and the Python programming language to create the system which he claims to have an accuracy rate of 95 percent.

The Braille to audio prototype happened to be Bhatt’s final year project for B.Tech which dates back to 2013. “Think of Braille as embossing,” Bhatt explained. “You have a 3×2 grid, so three rows and two columns wherein by touching that particular button, a person who’s not able to see would be able to understand that this is number 3… and so forth.”

This 3 x 2 grid formed the basis for the system Bhatt created that allowed calculations to be performed by the microcontroller. “A stepper motor would rotate which will again have the embossing and the person can touch the results and understand that this calculation has led to this particular output, so he’s well aware of what the number is,” explained Bhatt.

It was in 2019 that Bhatt got back to his college project again to further improve on the same using machine learning techniques. Bhatt said he had to create his own data sets using data augmentation given that he didn’t have access to available data. Once a substantial amount of data was created, Bhatt fed those to NVIDIA on Google Cloud. The real challenge for Bhatt was to replicate the same using machine learning and data science.

So, where do things stand at the moment? Bhatt said he is ready with the model but it’s such that only he has access to the tool that converts Braille to audio. Right now, what is pending is the development of an interface that can communicate with Google Assistant and the like. Bhatt however said he has tested the prototype with those with visual disabilities and the results have been satisfactory so far, with a correct audio translation generated most of the time.

Bhatt also said he intends to open-source the project after ha has developed an Android app on the same but didn’t reveal when is that expected to happen.

The post Indian data scientist creates tool to convert Braille to text and audio using AI first appeared on Good e-Reader.