My undergraduate final year project. I learnt Unreal Engine 5 from scratch to develop a image synthetic data generation pipeline. I then combined the synthetic data with the Cityscapes dataset to train a segmentation model on my university's remote GPU server, achieving a 6% increase in accuracy over the baseline model trained with only real data.
I led a team to develop a webpage capable of predicting OSA and its severity with an accuracy of 96%, effectively facilitating OSA diagnosis. I built the prediction models for the webpage using raw datasets collected from Malaysian hospitals which I cleaned and preprocessed.
