The Reserve Bank of India Governor, Shaktikanta Das, launched the Mobile Aided Note Identifier (MANI) app on January 1, 2020, to aid the visually impaired in identifying the denomination of Indian banknotes. The app is stated to be technological progression to various ways that enable the visually impaired Indians (color blind, partially sighted, or blind people) to identify and distinguish between the banknotes.
According to the World Health Organization (WHO), 20% of the world’s visually impaired population is in India. This is one of the significant reasons for RBI to empower nearly 8 million Indian citizens with an app that can help them to identify currency denomination and distinguish between them.
In this initiative by the Reserve Bank of India, Daffodil Software has been their technology partner who translated their vision into a mobile app. A leading software engineering company, based in Gurgaon, Daffodil software reveals that Artificial Intelligence is the underlying technology that allows the users to identify banknotes denomination.
One of the major challenges in achieving maximum accuracy was to identify the notes in different light conditions, from various holding angles. To sort this challenge, Machine Learning (ML) with Image Classification was identified as the most relevant technology for building the app. Machine Learning is an application of Artificial Intelligence that makes a system to learn and improve from past experiences.
For currency note identification, team Daffodil created a set of 1,50,000 images of old and new banknotes in the denomination of Rs 10, 20, 50, 100, 200, 500, and 2,000 that are in circulation today. Whenever a user scans a note, the app compares the scanned image of the note with 1,50,000 images in the data set. Image Classification technology, on the other hand, classifies the images according to its visual content and notifies the users.
One of the key requirements of the app was to make it work in offline mode to enable users to identify currency denominations in areas with poor internet connectivity. This functionality of the app posed challenges to the data set in self-training, i.e. as users scan more images in the app, the data set does not grow larger. Therefore, it was important for team Daffodil to intensively train the data set to give results with maximum accuracy.
Regarding the success of the MANI app, Yogesh Agarwal (CEO, Daffodil Software) says: “Daffodil Software is an early adopter of Artificial Intelligence technology which puts it into the leadership position to build an app like MANI. Being a government of India initiative, this was a key project for us as it was going to empower 8 million visually impaired people across India. For people with hearing and visual impairments, the app has predefined vibrations for different denominations of banknotes. For its technical and functional advances, the app has received an applaudable response from the users with 3,50,000+ downloads in the timespan of just one month.”
About Daffodil Software
For more than 20 years, Daffodil Software has been a trusted software technology partner to organizations across the globe. With our roots into innovation, tech agility, & time-proven processes, our team of 600+ technologists strives to shape the tech industry and help businesses elevate their value proposition through technology.