「初音ミク『マジカルミライ 2025』」プログラミング・コンテスト応募作品です。
Produced for Hatsune Miku "Magical Mirai 2025" Programming Contest.

Over the past 2 years, I started to develop a small obsession with Vocaloid and decided to take part in this contest. I was super excted when I learnt about the contest and already started working on my submission in November 2024. The competition required the use of the TextAlive App API to make a lyric app/game. So I decided to make something interesting, and integrated on-device markerless motion capture into a lyric game. Using Google MediaPipe's Pose Landmark Detection Library, the program tracks the location of the users hand to detect when the user taps on the lyrics which appear on the screen. I needed to use the full pose landmarker because the hand landmarker detection did not work when moving hands quickly. I think that this is one of my coolest projects. I was disappointed when I realised that I did not get pass the judging stage for public voting, but I am still elated that I was able to take part in this contest. I realised that even though my project was cool and innovative, it severely lacked better graphics and images which was one of the key components in the judging criteria. I should have worked with other people who knew how to draw. Furthermore, the lack of experience with similar rhythm games made the project less successful. (At this point of writing, the only rhythm game I have played is this game that I have developed, not even Project Sekai. I'll probably try one day when I have the time.)
Overall, I really enjoyed programming this game and will remember this unforgettable experience forever. It holds a special place in my heart even though its not my most successful one.

Check out the project's GitHub repository. You can also play the game here.

Screenshot of scrcpyMediaController in Hyprland

Screenshot of scrcpyMediaController in Hyprland

Screenshot of scrcpyMediaController in Hyprland

Previous Next

I wrote this script when I got tired studying for Organic Chemistry in preparation for my GCE O-Level examination. Slowly, I kept fixing some bugs until I was satisfied with it. I might compile a proper release binary or maybe even learn C and make a pull request to scrcpy after my exams.
Click here to see the project in GitHub.

Introducing the the Integrated Classroom Hub (ICH), an all in one classroom manager.
The ICH fixes multiple ICT problems in my classroom such as poor WIFI connection that constantly disconnects. Using a spare Singtel Mesh Router I had lying around in my house, the ICH auto log in to the school WIFI and start a hotspot which goes out through the ethernet port of my laptop to the Singtel Mesh Router. The ICH is also supposed to come with a better web filtering system but it was never completed before Secondary 4. The proposed web filtering system uses a transparent proxy to monitor and block every single outgoing request to the internet from the students. This is better as the school would only need to maintain the ICH in each classroom rather than doing software checks on every students' laptop every semester. The ICH also starts 2 Python Flask Servers. The first one running at port 80 handles shortcuts. When connected to the Wifi, if you type igc/ in a chrome browser, it will redirect you to Google Classroom (*IGC stands for ICON Google Classroom). It does this by DNS Hijacking the domain igc which works despite not being a valid domain name. The second one running at port 443 is a website that manages reminders, important dates, homework and overdue homework for my class. The website also comes with various tools such as an image to text converter and WebRTC Screen Sharing to screen mirror (Cast) the students Chromebook to the ICH (Which is also connected to the projector. DNS Hijacking also meant that I could get any domain I wanted for the website, except that the website would only be accessible when connected to the ICH's WIFI.

A Demo of ScreenShare from the Receiving end. When a YouTube Video is played on the client, there is smooth playback.

ScreenShare is a simple browser to browser screen sharing application between 2 devices. It was initially intended to help my classmates and my Math teacher screen mirror our Chromebook and her iPad to the Projector but I later discovered that WebRTC (The technology used by ScreenShare) does not work on iPads. It then became a prototype for the Integrated Classroom Hub (Another prototype that I will talk about later). You can see the project and try it out yourself here.

My program output (Trackers)
I have obtained permission from the people in this video to put this GIF here.

My program output (Annotations)
I have obtained permission from the people in this video to put this GIF here.

A failed test run of my program when under testing. The final generated graph is upside down.

A test run of the software I wrote for face data collection. Many frames in the screen recording were dropped due to lack of CPU resources.
Take not that this is a screen recording of 2 displays side by side.

Previous Next

Using my knowledge on Python, I applied it to my Math Modeling Projects. The first one was a Beauty Rating Software. It determined which of my classmates were the most beautiful based on the golden ratio. You can find more about it here along with the code written for the project. The second one, which I did not upload as the sample data would contain most of my classmates faces, measured the angle in which the basketball was thrown at as well as create an quadratic equation to show the trajectory of the basket ball thrown, using a sample video. These projects gave me the chance to experiment with Artificial Intelligence using Python and allowed me to play around with various python packages such as torchvision, matplotlib, numpy and opencv-contrib-python.
*For the first project, when I was training the Facial Detection Model on my laptop, I was not aware that there was already a trained model (.pt file) in the project's repository.