Portfolio Entry

This is a brief summary of FYP 1 for the semester

During the earlier weeks of this semester, our supervisor, Dr. Mohd Azhar bin Abdul Razak encouraged us to do some exploration regarding the research title for the FYP. The process was personally challenging as I was not knowledgeable regarding the project or research that has been done in the past as I am not involving myself in senior FYP presentations.

For starters, I visit the open source website IEEExplore in hope of finding some inspiration regarding any topic for my title. Luckily, the website contains a bar chart exhibiting the number of research points of each topic. After skimming through the chart, I was interested in exploring machine vision. I felt it was the right topic for me since recently, I attended a talk related to machine vision.

Even though I was determined to do research on machine vision, I still have no knowledge at all on how it works. Then, I continued the exploration in order to see what scope of research machine learning had. With no prior knowledge on doing literature review, I spend a huge portion of my time reading every paragraph of the research paper that I picked. Furthermore, not knowing what the paper was all about, I spent more time confusing myself reading that single paper. At the end of the day, there was almost no progress in my knowledge of the scope of the study.

In order to do some discussion on the research field, I went to YouTube for inspiration and I found a scope of my research. The scope was deep learning. From what I have watched on YouTube, I was actually the tip of an iceberg. I did not know there are still more under deep learning yet there is a small progress of my FYP. At that point of time, I felt like I was behind everybody else in terms of progress.

Then, I had some discussion with my supervisor with little knowledge of machine vision and shallow knowledge about deep learning. After discussion, he helped me with my title of FYP which was "Gesture Recognition for Smart Home using Deep Learning".

Next, I tried my best to do my literature review but I did not realize that the title contained the scope of my research. Thus, the process was literally to research papers that were widely related to gesture recognition. Even though I went through the papers I did not try to understand it. It turns out it was a mistake as I was spending my time with research that actually was outside of my scope.

Fortunately, I am able to prepare the FYP presentation slides and I did realize that my research seems still undecided, ironically. After presenting my research to the panels, I just realized that my objective was not firm after receiving comments from the panel. The realization came through and  I went back to the literature review phase, and it turned out I am able to find research papers that are similar as the scope was actually the same as mine. 

Because of that, I am able to revise my research and the content seemed more updated and decided on what to do. Specifically, I will do research on static hand figures using CNN for smart home automation. I still need to know about how to obtain the data sets- whether to make my own, or transfer learning, or find existing datasets. What to do if we use the camera for input, as I understand that the input is a picture. Do I need to do segmentation and background subtraction?. Lastly, I never implement any IoT device but I am planning to work with Arduino.

In conclusion, my methodology is still not firm and I need to do more research regarding those matters. Moreover, I need to ask my supervisor's guidance more often as it is backfiring.