Yi-Hsin Li - early stage researcher within visualization | miun.se

Yi-Hsin Li - early stage researcher within visualization

Mon 24 Oct 2022 13:04

Yi-Hsin Li works as early stage researcher at the Technische Universität Berlin, Germany, and Mid Sweden University. Her research topic is “Gated Experts Compression of Light Field Video“.

Yi-Hsin Li
Yi-Hsin Li PhD-Student within Plenoptima-project at Technische Universität Berlin, Germany and Mid Sweden University

Tell me about yourself, your background?

I come from Taiwan, a country with freedom. I studied electrical and computer engineering for my BA at National Chiao Tung University in Taiwan and during this time I developed a particular interest in computer vision and machine learning technology. Furthering this interest, I undertook an MSc in electrical engineering and computer science at National Taiwan University. My MSc thesis was about developing deep learning techniques to combat the effect of fisheye image distortion on face recognition. After finishing my master’s degree, I worked as a research assistant in a co-operative project for my supervisor’s group on medical image analysis. During this time, I develop a GAN model to convert optical coherence tomography (OCT) images (a kind of gray-level images) to Tuberculosis (TB) images (a kind of stained images) with my knowledge of the image analysis and my experience with deep learning techniques.

Why did you apply to the Plenoptima project?

First, I am interested in image processing and machine learning. Actually, I have great experience in this field. Second, I’m interested in real-time sports streaming. Specifically, truly immersive experiences of scenes captured by omnidirectional devices. In my opinion, to achieve real-time, compression is the most important issue. Therefore, I applied for the ESR9 position, which is about compression for light field data.

What have you done so far as an PhD-student?

The main problem in my research topic is that the optimization processing for encoding LF is still offline. Therefore, I have developed an initialization method for placing kernels in good initialized locations so that the remaining optimization processing would be quick and the performance of decoding results is improved.

What are your thoughts about the upcoming visit as a PhD-student at Mid Sweden University?

I am looking forward to working at Mid Sweden University. I attended a training school and workshop in September. The environment of the University as well as the colleagues are fantastic. Furthermore, it will be my time to work in extremely cold winter (at least for me), and I believe that it will be a quiet interesting experience for me.

What kind of expectations do you have for your research project and the upcoming network training?

I expect to be inspired by ideas coming from all ESRs and discuss the problems we face, how we tackle them and compare the solution to see the advantages and disadvantages of all the solutions. I also hope to learn about the state-of-the-art techniques in the fields of other ESRs.


Yi-Hsin Li

Doktorand|Doctoral Student


The page was updated 10/24/2022