Interview with Professor AG Ramakrishnan Department of Electrical Engineering, IISc

2015.06.08

 

In this interview, we feature Professor AG Ramakrishnan and research work carried out by him and his research students on Medical Intelligence and Language Engineering (MILE) lab, at Indian Institute of Science (IISc). His work on Optical Character Recognition (OCR) technologies applied for visually challenged has been recognized with Gandhian Young Technology Innovation (GYTI) Award.

  1. Please share with us good work that IISc EE MILE lab has been doing for the cause of access and inclusion for the visually challenged.

Rcmct Worth Trust Rehabilitation Centre, Chennai has been using our Tamil OCR for the past five years for converting printed school and college books to e-text and Braille books. We have been working with them closely and systematically, and based on their requirements and feedback, we have continuously improved our performance and user interface. Now, they have converted over 500 books to Braille books, which are being used by hundreds of visually challenged students. Worth Trust has also digitized quite a number of Tamil books for Bookshare.org, a nonprofit company in USA, funded by US Government and serving the blind people. We received the Manthan Award 2014 for e-accessibility and inclusion for this work. Recently, Karna Vidya Technology Centre, Chennai has also started using our ShriVeRa OCR GUI with our Mozhi Vallan Tamil OCR. One can see how our OCR technology is being used by the visually challenged in this video on youtube:https://www.youtube.com/watch?v=pMGi__btTOc

Mitra Jyothi and Braille Transcription Centre of Canara Bank Relief and Welfare Society in Bangalore and SAVK, Shimoga are organisations in Karnataka that are using our Lipi Gnani Kannada OCR for creating Braille books.

Anna Centenary Library, Chennai has been using our web demo of TTS to convert short messages to speech and sending them to their visually challenged members, who are about thousand in number. Recently, we have successfully made our Tamil and Kannada TTS, SAPI compatible and a few blind students have started using our TTS on NVDA screen reader platform to access the e-text created by our Kannada OCR.

2. GYTI award for work done by you and team is a great acknowledgement. It would be nice to know the challenges you encountered and milestones accomplished in this journey.

March 8, 2015, Rashtrapati Bhavan; Prof AG Ramakrishnan and HR Shivakumar (PhD Student) on extreme right receiving GYTI award.

It is a long journey. We started working on OCR in 1999, about 16 years ago. For quite some time till we proved our commitment, obtaining sufficient funds for research and development was extremely difficult. I had to work for months and pursue relentlessly to obtain a funding of one lakh or a few lakhs.

 

The salaries we were allowed to pay the project staff were highly limited till very recently. Because of this, it was almost impossible to get experienced engineers or programmers to work on the funded projects. Only fresh engineering graduates join the project, and most times, there is a long learning curve. Then, most often, when I think the person has learnt a lot and I am going to get more of quality work from him/her, he/she gets admission abroad for higher studies and says goodbye! I am not blaming them. Given the paltry salaries and no direct provision to register for any research programs in IISc even on part-time basis, it is inevitable and expected. One can get some idea of these challenges in my interview by The Mint on the occasion of Manthan award: https://www.youtube.com/watch?v=IlA-0WmGtQI

 

Dec 4, 2014: India Habitat Center; Prof AG Ramakrishnan (Extreme right) and HR Shivakumar (PhD Student) receiving Manthan Award.

We found our first committed partner in Worth Trust in 2010, and to begin with, we gave them the C version of Tamil OCR on Linux platform funded by Technology Development for Indian Languages (TDIL) under the Ministry of Information and Communication Technology. Since the user community was not at all familiar with Linux platform and the OCR had multiple dependencies, we decided to come up with a completely new design of an OCR, totally free from any external dependencies in terms of ideas and libraries. My research student, Mr. Shiva Kumar, developed a fast, Java based OCR, where even basic image processing operations were coded by us, without making use of any external code or libraries. Thus, compared to the 5 minutes conversion time per page of our TDIL OCR, our own new Java OCR performs in 2 seconds flat. This was augmented by a custom designed user interface, named ShriVeRa GUI, which facilitates conversion of a whole book of a hundred pages in a few minutes. A brief video on the versatility of our GUI can be seen in this video on youtube: https://www.youtube.com/watch?v=zRRr856jA-4

This work was what was rewarded by the GYTI award.

3. Please share with us what keeps you personally excited on this journey. Please share your personal story and some anecdotes that you fondly recall.

It gives one an extraordinary satisfaction that one’s life has been useful to somebody. As an engineer, it gives me enormous fulfilment, when something I worked on is being used by one or more people on a regular basis. Early in my career, I developed an EEG mapping software, which ultimately reached at least a hundred neurologists and it made me so happy. Then I developed a heart rate variability analysis package, compatible with the International standards at that time, and it was used for research in NIMHANS, AIIMS, Institute for Aviation Medicine, Swami Vivekananda Yoga Anusandhana Samsthana and a hospital in Mumbai.

Kaushik Mahata, a Bengali student, was the first to work on Tamil OCR. It is a very interesting story. He was supposed to go to University of Kaiserslautern in Germany on an exchange program to work on medical images. For some reason, he was denied a visa. He got so upset that he told me that not only he did not want to go to Germany, but also that he would not like to work on that project at all. He was ready to take up any other project. I asked him whether he would work on Tamil OCR and he said yes. That is the first time I started work on OCR and I was also learning to design an OCR! He did an excellent job and during his final project viva, when a Professor from the Department of Computer Science asked him how he could work on Tamil, being a Bengali, he surprised him by reading the Tamil text on his slide. Such was his commitment that he had taught himself Tamil during his project months. So, occasionally we get such extraordinary students in IISc, who inspire us! It is a blessing!

Another interesting story is how I got Shiva Kumar as my true partner in all my endeavours. Our first encounter was in Landmark Education, where I was leading a seminar on a voluntary basis and he baffled me when he said that my smile is artificial. I thought that others may not think so and I asked the audience (about a hundred of them) who else also thought so. About 20 hands went up and I was really taken aback! It is another thing that I saved my face by asking the audience once again as to how many people thought it was genuine and a slightly larger number raised their hands. A few years later, he called me over phone and said that he wanted to know whether he could pursue PhD under my guidance. By that time, he had also become a voluntary leader in Landmark Education. However, being in Electrical Engineering Department, I was not sure of guiding a computer science graduate and hence, I told him that we could meet and discuss. When we met, he instantly reflected my commitment to create language technologies for empowering and enabling the visually challenged. Today, I owe the reliability and performance of my OCR and TTS to him. He also brought into our lab a lot of industry standard, software engineering practices, for which I am ever indebted.

4. Would you like to share key developments in corporate, government, regulation and other ecosystem that are promoting accessibility and inclusion?

Recently, the NDA government has revised the Corporate Social Responsibility (CSR) norms. As per this, companies having at least Rs 5 crore net profit, or Rs 1,000 crore turnover or Rs 500 crore net worth need to spend 2% of their 3-year average annual net profit on CSR activities in each financial year. Some corporates are now looking at accessibility and inclusion for their CSR compliance. Further, the Government has also passed the rule that any organization can digitize copyrighted books solely for the purpose of serving the visually disabled. These two policies will go a long way towards increasing accessibility for the disabled. I have also suggested to a major IT company that they can create a server for maintaining the digitized e-text of school and college books, interlink all the NGOs serving the cause of the blind and thus dramatically expand the availability of Braille and DAISY books. Once this is implemented, all duplication efforts can be stopped and the productivity of the various NGOs in increasing accessibility of books to the visually disabled will skyrocket.

5. How has technology helped advance the cause of accessibility and inclusion? How do you plan to make the technologies you own available to society at large and the needy?

Before the internet era, people could access knowledge only through Braille books or human volunteer readers. People had to painstakingly type the entire text and then emboss it as a Braille book. Because of lack of systematic and easy communication, I am sure the same books were created in Braille in multiple places, by repeating the entire human effort involved. Today, by using optical character recognizers (OCR), for example, our Mozhi Vallan Tamil OCR or Lipi Gnani Kannada OCR, any book can be converted into e-text or Braille embosser compatible codes in a matter of hours. By creating a well-coordinated e-group of the voluntary organizations serving the cause of the visually disabled, unnecessary duplication of effort can be totally avoided and the created e-text or Braille codes can be shared across the institutions. The e-books can be listened to, using screen readers or using mobile DAISY players, which provide random access to the contents of a book. These DAISY players give access to even people, who don’t know Braille or cannot use a computer.

We intend to start a company soon to develop our technologies for further improving the performance and the user interface. The Indian Institute of Science has created the Society for Innovation and Development, through which the Institute will have certain percentage of shares in such endeavours by the faculty. We also want to expand the scope of our text to speech and OCR engines to other languages. We also have other technologies such as handwriting recognition in Indian languages, which also will be converted to useful products. To the extent possible, we would give free access to our technologies for the needy and charge people, who can afford and also the corporate customers for other profitable applications of our technologies. For example, the OCRs are regularly used around the world for transforming traditional printed material libraries into digital and online libraries. Text to speech technologies are used for announcement and enquiry systems, besides multimedia education systems.

6. What role you see Getmeenabled.com playing in the ecosystem of disability?

Getmeenabled.com is providing a greatly needed service of making available all the available tools and technologies that enable and empower people with disability. It is a one-stop shop (solution) for everything they need to enrich their lives and is a remarkable contribution. You can additionally give a list of all the NGOs and other organisations working for the PWDs, and may be their contact addresses or website links.

Back