Drishti: An Acute Insight into AI
Standing across the counter, all I can hear is the receipt printer brrh, click, clatter, clunk, and ka-ching!
‘It’s a thousand and fifty bucks sir!’ the cashier says. I reach for my back pocket and feel the first note that comes out. “A single fold for a thousand and a twisted fold for a fifty”, I repeat to myself. However, my notes felt unusual to my hands today. They seemed like they were folded from the side, maybe a little bit twisted too, I couldn’t make out otherwise. “Oh, not again!” I thought to myself as I remembered being in a hurry the last time I folded them. So I turned around to check if anyone could tell how much money I had in my hands. The cashier hastily tells me to do my business a bit quicker. Feeling helpless, I keep all the money I have on the counter and tell her to take the required money and give back the remaining.
This is a glimpse of the daily struggle of a visually impaired individual while paying for his grocery. There are about 95,000 such visually impaired individuals in Nepal (2011 census) who go through the same on a daily basis. This figure has escalated over the years which truly validates a need for some kind of solution to the daily challenges visually impaired people face due to their impediment. One such prominent area is their inability to actively participate in the economy through financial transactions.
Thanks to advancements in artificial intelligence, there are a few mobile applications working on improving the status quo. These applications can be installed on smartphones and can help identify scanned currency notes via the camera.
However, these applications are still a work in progress. Many of them often have inaccuracies and language barriers for the end-users.
Hoping to further bridge the gap between these existing advanced technologies and our end users, we are introducing Project Drishti.
Project Drishti is driven by a team of like-minded individuals together with the amazing powers of Artificial Intelligence and machine learning working to empower low vision and visually impaired people in Nepal.
We carefully studied the complications in the pre-existing applications and came up with solutions to those problems and incorporated them into the Drishti App.
We are building a computer vision model for cash recognition; an android based mobile application that helps blind individuals scan currency bills and get audio output.
The salient features of which are listed below.
- Our model is designed to detect Nepalese currencies and interfaces with the end-users in Nepali.
- We have also leveraged the native OS accessibility features to provide helpful capabilities such as screen reading and haptic feedback.
- We are committed to providing a smooth user experience to our visually challenged users and hence refrained from any kind of monetization of the application through advertisements.
- It serves its purposes without the need for an internet connection. The model works locally, the data has little to no chance of getting breached and works in real-time.
- Our application stores the scanned readings of the bills.
- The history feature allows you to find out the day and the time you scanned a particular note as well as the total amount you spent on that particular day, week, or month.
For our product to reach our end users, our team will be collaborating with the Nepalese Blind Youth Association. The technology department of the association which works in providing technical assistance to the sight-impaired, will test our product and later distribute our application to blind individuals after quality assurance. This will help us gain the trust of our users and receive honest feedback on the things that need to be improved.
Nepali Currency Detection is just a small part of the Drishti App.
Upon talking to our end users from the Blind Youth Association of Nepal, we have come to realize that there is also a dire need for features like performing short Nepali text and document readings along with cash recognition.
We aim to make Drishti a full-fledged application to help the blind/low vision people in all of their day-to-day activities and for that, we will need your help.
While we work on strengthening the application, we have also decided to open-source our project.
We encourage any interested individuals to view our project source code in Github (link given below) and contribute to adding new features to our application and help us achieve our goal of empowering the low vision/blind individuals.
We also have a project document and a technical design document to help onboard new contributors to the project. You can make changes, add more features as you please, and directly send us a pull request that will be merged after approval. If you have any queries and concerns, feel free to contact us at email@example.com. Please feel free to forward this article and the GitHub repository link given below to someone who might be interested in developing an AI application to solve a real user need!
If you’re not familiar with coding or machine learning, you can still support Project Drishti!
You can download the app from the link given below and give us feedback. And if you find any bugs, please let us know through our social media pages or website, or file an issue on Github and our engineers will be on their toes!
We wish to be the platform where we inspire innovators and learners trying to bring an impactful change in their community through their skills and works.
Have a LOOK into this VISIONARY project and you shall be BLINDed by the world full of possibilities!
The link to our Github repository:
Download the app on play store: