About Me

I am currently a Master Student in the Department of Electrical Engineering at National Tsing Hua University (NTHU), under the supervision of professor Min Sun. Additionally I am serving as a deep learning intern at MediaTek Research. I am proficient in several programming languages, including Python, C/C++, HTML/CSS, JavaScript, MySQL, Matlab, and Shell Script. Additionally, I am experienced with frameworks such as Linux, Git, GitHub, PyTorch, and LATEX.

My research at NTHU focus on data augmentation on panoramic images. I try to solve the scarcity issue of panoramic datas by implementing a brand-new data augmentation that not only fit panoramic images but also enhances dataset diversity. My new paper on this work was accepted by BMVC 2023.

As a deep learning intern at MediaTek Research, I was deeply involved in crafting applications that harness accessibility services within the Android platform and LLM. I was also dedicated to advancing an expansive Chinese language model alongside its related tasks. Furthermore, I collaborated with colleagues from the Cambridge research team to explore AI-driven methodologies for generating test cases in ASIC design verification.

As a machine learning intern at Roku, the leading streaming TV platform in America, I use AI to help improve picture quality setting based on different scenes. To be more specific, I generated a large set of images with new classes from Roku's internal dataset and trained and fine-tuned MobileNetV3 on the new data for scene classification. Additionally, I deployed the model on Roku TV to enable real-time content classification, with dynamic picture quality adjustments (contrast, clarity, color) based on the detected scenes.

Publication

Project

  • Computer Vision: Real-time Fighting Game

    Develop a two-player fighting game using real-time human pose estimation for avatar control through poses and Utilize GAN-based face morphing for avatars to shift between different looks smoothly.
    Code

  • LeetCode

    Solve over 1000 leetcode problems utilizing Python or C/C++, spanning various problem types such as dynamic programming, backtracking, binary search, BFS, DFS, greedy algorithms, monotonic stack, and so on. The solutions for both the latest biweekly and weekly contests are also available here.
    Website Code

Education

Experience

  • Deep Learning and Software Intern in MediaTek Research, Taipei, Taiwan

    Dec. 2023 - Apr. 2024

  • Machine Learning Intern in Roku, Hsinchu, Taiwan

    May. 2024 - Aug. 2024

Others