News

Providing Big Data-Based Customized AI Solutions to Governments and Companies

2019-09-09 Academic

Student Jihyun Kim (Ceramic Arts, ’21) has established Korea Deep Learning, which provides customized AI solutions based on big data to governments and companies

Only one month after its founding, Korea Deep Learning signed a contract with the Korea Post to provide its logistics optical character reader (OCR) system. There are some adjectives that can be attributed to Kim: 23-years-old, female, and non-major, which create the worst conditions for a person to work in the field of artificial intelligence. Whenever people cast a suspicious eye, saying “What on earth can you do?” she answers that her lack of experience led her to accept the stories of not only success but also failure from various fields and, in turn, gave her a broader perspective. During an interview at the Gwanggyo Business Center, Kim pointed out that being a pioneer empowers her to grow.

Her AI solutions use deep learning and machine learning
Q. What kind of company is Korea Deep Learning?
A. Korea Deep Learning provides governments and companies with customized solutions based on a type of artificial intelligence that processes patterns, recognizing faces, texts, and objects and analyzing images. The core technologies are self-developed deep-learning and machine-learning prediction engines. The representative item is the Logistics OCR System. Currently, the barcode is used to recognize an address and classify mails by the address. However, the false recognition rate of the barcode-only recognition system reaches two percent. Although the number sounds small, the number of letters and packages processed at distribution centers or mail centers is so large that two percent means 200,000 to 400,000 unsorted postal deliveries.

As a solution, we developed a logistics scanning system that uses artificial intelligence to recognize not only barcodes but also the addresses on packing slips. It recognizes addresses on the fast-moving conveyor belt within 0.3 seconds, minimizing the number of unsorted logistics, and the recognition accuracy is as high as 99.8 to 99.9 percent. Currently, it is being used to build a system at our post office logistics center, and, if all goes well, we plan to expand supply.

Q. Any items other than the OCR system, that you are working on now?
A. In addition to text recognition, I am developing artificial intelligence solutions in image analysis and image processing. For example, I am planning to build a system that collects data on the traffic volume of consumers visiting a retail store, their preferred items, and which items they looked at and considered purchase, and then we propose a management method for the retailer on the basis of the data. Also, in collaboration with veterinarians and people in animal husbandry, I am working on developing a smart farm solution that will be used as a preventive measure for epidemics.

Corporate goals are converging selected data and creating synergy
Q. It is refreshing to see a ceramic art student in the field of artificial intelligence. What was your motivation?
A. Since my parents are developers and entrepreneurs, I was even as a child more interested in starting a business than being an employee. However, it was not until I took a class called Startup and Challenge that I decided to work on it in earnest. Taking advantage of my major, I used diatomite, a soil that absorbs and holds water, to make an umbrella hanger. This item gave my confidence a quick boost by earning the Grand Prize at the Kyung Hee University Creative Idea and Design Competition.

Participating in the KHU Valley Program helped me realize my business items. In addition, as a grand prize winner, I had a chance to work as an intern at the Vietnamese branch office of N15, a hardware accelerator. By winning various startup competitions, I could firmly establish a framework for my startup while raising capital. In the meantime, however, a big company released the same product. It taught me the limitations of manufacturing startups without capital and a production base, and that was when I decided to open a software business instead.

Q. What do you plan after this?
A. The solution business is directly connected to the actual issues of governments and companies. Developing a solution creates an idiosyncratic and definitively validated business model. Similarly, the data generated here is high-quality data specialized to solve problems in a specific field. Then, I want to converge the selected data and use it as the basis for yet another solution, to generate synergy.

  • News List

    • Kyung Hee’s Sma...

      Kyung Hee’s Smart Tourism...

      2019-11-29 Academic

      Read More
    • A Sweet Cake De...

      A Sweet Cake Delivering K...

      2019-11-29 Academic

      Read More
  • Highlight News

    • Professor Juliu...

      Professor Julius-Jeongwon...

      2012-06-05 Academic

      Read More
    • Professor Sang-...

      Professor Sang-Ho Lee to ...

      2013-11-22 Research

      Read More