Shengkai Xu

Hello this is Sam (Shengkai)! I am a first-year PhD student under the guidance of Dr. Hongfei Xue, Dr. Pu Wang, and Dr. Ahmed Helmy at the University of North Carolina at Charlotte. My research focuses on computer vision utilizing deep learning models for health applications, with a particular emphasis on human pose estimation on edge devices and processing human pose estimation using mmWave devices detection methods. You can find my work on my GitHub page.

I have a strong background in hardware and firmware engineering, particularly in embedded systems and IoT devices. In my free time, I enjoy tinkering with electronics and exploring various engineering challenges, such as motorsport and automobile repair. Additionally, I love outdoor activities like rock climbing and camping, which allow me to engage with the world beyond technology.

Connect with me on LinkedIn.

Education

M.Sc. Computer Engineering | University of North Carolina at Charlotte

Research in Deep Learning and Computer Vision on Edge Devices/Servers

B.Sc. Computer Engineering | University of North Carolina at Charlotte (Jan 2018 - Dec 2022)

Dean's List
Concentration in Machine Learning
Minor in Mathematics

Skills

Category	Skills
Programming	Python, SQL, Swift(Xcode), Embedded C, VHDL
ML/AI	Pytorch, TensorFlow, Keras, Panda, scikit-learn
Embedded System	ESP32, STM32, Arduino, RaspberryPi
CAD	SOLIDWORKS, EAGLE, KiCAD
Cloud Service	AWS, Google Cloud, OpenAI, RunPod
Misc	Linux, Git, GitHub, Vivado, LaTex

Experience

Research Assistant | Dr. Hongfei Xue (Jan 2024 - Present)

Implemented the deep learning model on edge devices, such as leveraging the iOS Neural Engine to perform 3D human pose estimation using 2D images as input with optimize performance and efficiency. Focused on enhancing real-time processing capabilities for applications in health and fitness monitoring.
Utilized mmWave sensing technology to develop a system for classifying human activities. Implemented advanced signal processing techniques to accurately detect and categorize various human movements. Focused on enhancing the accuracy and reliability of activity classification in real-time applications.
Creating and implementing a ground truth model using RGB and Depth Cameras for human pose estimation.

Research Assistant | TeCSAR (Aug 2022 - Jan 2023)

Developed AI pipeline for civilian security and public safety systems.
Achieved a 50% increase in processing speed.

Embedded Firmware Engineer | Oxit (May 2021 - May 2023)

Led R&D initiatives for IoT service and device automation.
Developed modular libraries for embedded applications.

Publications

S. Xu, M. N. Hasan, R. E. Thomas, A. Lopez, and S. Shue, "Charlotte Area Traffic Light Dataset," in IEEE HONET 2023, 2023. In Press.

A. Brooks, B. Bryant, C. Spoerer, M. Lust, S. S. Xu, Y. Li, N. N. BouSaba, and D. Maity, "Path Planning for Robotic Delivery Systems," in SoutheastCon 2022, pp. 421-426, 2022. doi: 10.1109/SoutheastCon 48659. 2022.9764058.