My First R&D project as an Intern at 99x

A Research and Development project that I worked on as an intern at 99x.

Research and development is one of my favourite areas to engage in since you get to perform the research and work on the implementation yourself, which means you will have the greatest idea of the project you execute, and the problem solved by the project.

During my second year of my bachelor’s degree, I completed another R&D project before joining 99x as an intern; nevertheless, this article will focus on the R&D I was assigned to work on during my internship time.

My first task for the project, Norkart, to which I was assigned as an intern was an R&D task. This was a new feature that would be implemented into the application we were actively working on. I was excited since it would be my research and I would be able to learn something new simultaneously.

The issue I had to tackle had to do with extracting meter readings and serial numbers from photographs of water meters. This necessitates data science, and I was pleased since it is in my field of interest, so I will be more focused on my passion.

I knew that OCR (Optical Character Recognition) would be utilized in this case scenario, but I still researched on Google for alternative ways for comparable problems like these to be addressed, and it was certain that OCR was the expected approach for this situation.

Later, I researched many platforms that provide this OCR service, and I was able to identify the four best approaches to implement OCR, which included Vision AI from Google, Computer vision from Azure Cognitive Services, and the following two libraries, pytesseract and keras-OCR.

I then utilized all four services, observing how they performed in terms of accuracy, speed, reliability, file size supported used, cost, and other factors. The reason for implementing all four of them was to determine which one was the best long-term solution for the project, thus all of my findings were reviewed with my team lead as well as the client.

Since Azure computer vision was ranked first among the four services, I proceeded to create the functionality using test water meter images provided by the client. Although this was my first dive into the area of azure AI, I was fortunate to be able to work with Azure Cognitive Services and learn from it. Moving on, I was able to construct a working API service for this problem and test it; thus far, it has behaved very well, and I was able to demonstrate it to the clients during meetings.

So this was my first R&D project as a software engineering intern, and I’m looking forward to more in the future.

More content at plainenglish.io