Newswise — A team from Chula’s Faculty of Engineering have made use of AI Deep Tech to develop a program that scans documents and images into OCR documents. The program is more than 90% accurate when reading Thai scripts and Chula’s UTC is now ready for a spin-off to the market through Eikonnex AI Co. Ltd.
As many of us who work with information on paper documents such as questionnaires and research reports know too well that the most tedious and difficult part of each job has to do with transferring all the answers or data to digital files for storage by typing them page by page. This task is extremely time-consuming, straining your eyesight and risking body aches and pains that come with what we now call Office Syndrome.
For this type of work, Associate Professor Dr. Thanarat Chalidabhongse of the Department of Computer Engineering, Faculty of Engineering, Chulalongkorn University can replace that laborious effort with the use of AI: Artificial Intelligence. Along with Dr. Thananop Kobchaisawat and a team of doctoral students, Dr. Thanarat took on the task of developing “ReadMe”, an OCR (Optical Character Recognition) program that scans texts in both Thai and English on documents, images, or video files and converts them into a digital script.
What is OCR?
OCR or Optical Character Recognition technology is a program that digitizes texts from images or videos. The data can then be applied and used right away. Two types of images can be scanned: Scene text image and Document scanned image.
OCR technology can be applied for use in various ways such as reading postal codes on envelopes for automatic sorting of mail and reading the numbers on wheelsets to determine the position of the train’s carriage at a certain time. It can also be used for car cameras to read traffic and direction signs and or to help read various signs for the visually impaired.
For Dr. Thanarat “OCR is an AI innovation and a time-saving technology that helps lessen the use of human labor. It is convenient and significantly helps facilitate our daily lives.”
ReadMe reads Thai with the fluency of a native speaker
The accuracy of the OCR technology has greatly improved as a result of constant developments over the years. One of its weak points that was an obstacle for its Thai users had to do with its ability to read the “Thai language.”
“The grammatical structure of English makes it easier for a computer to read. Thai, on the other hand, is far more difficult with numerous letters in the alphabet, vowels, and tonal signs. In one line there can be as many as four levels of scripts while in English there is only one.” Dr. Thanarat explained.
Fortunately, such problems have now been eradicated with the use of Deep Learning technology that enhances AI intelligence.
“Once the new technology has been introduced, it can increase the accuracy so that the OCR can read Thai with a level of fluency that is comparable to a Thai person reading it.”
Read Me also lessens the burdens in the business sector
ReadMe is the first product of the Eikonnex AI Co. Ltd. with the support of the University Technology Center (Chula UTC). Dr. Thanarat and her team developed this product after a business survey had been conducted.
“A lot of the work in the business sector involves documents which still rely on a lot of manpower in filling out data. ReadMe was produced to help the business sector. Emphasis is placed on accurate reading to ensure the least possible mistakes.” Dr. Thanarat explained citing the use of ReadMe in the banking business.
“Agencies that have used ReadMe have found that in comparison with OCRs from other companies both Thai and foreign, ReadMe is the most accurate with up to 92.6% accuracy and can certainly help avoid instances of human error.”
At present, Eikonnex AI offers ReadMe services through customized software development or licensing for use with the agency’s own applications.
“We are now in the age of digital transformation. All agencies need to adapt and digitize everything. ReadMe is a tool to facilitate this digital transformation with relative ease.”
Will the use of ReadMe cost manual workers their jobs? Does the AI “replace” or “facilitate” things for us? How should human beings deal with these changes and situations?
For this Dr, Thanarat left us with these words of caution, “technology is moving very fast. OCR or ReadMe are definitely on their way. Many jobs will inevitably be replaced by technology which means that human beings will need to adjust to the situation and take on the tasks that computers are still not able to perform.”
More information about ReadMe can be found at https://www.eikonnex.ai/ A sample version of ReadMe is also available at https://readme.eikonnex.ai/.