CV

Yucheng Zhang

Email: yz90@illinois.edu | Mob: +1 2177214643

EDUCATION

University of Illinois Urbana-Champaign, Urbana, IL, USA

Bachelor of Science in Computer Engineering (expected July 2025)
Sep. 2021 - Present

• Cumulative GPA: 3.97/4.0
• Core Courses: Introduction to Algorithms & Models of Computers, Computer Systems Engineering, Probability with Engineering Applications, Machine Learning, IoT and Cognitive Computing

Zhejiang University - University of Illinois Urbana-Champaign Institute, Zhejiang, China

Bachelor of Engineering in Electrical and Computer Engineering (expected July 2025)
Sep. 2021 - Present

• Cumulative GPA: 3.90/4.0
• Core Courses: Data Structures, Computer Systems & Programming, Numerical Analysis, Applied Parallel Programming, Data Science Analytics using Probabilistic Graph Models

AWARDS AND HONORS

• Gold Medal, 46th ICPC Asia Regional Contest Jinan Site (Top 10%) (2021)
• Silver Medal, 2021 China Collegiate Programming Contest, Weihai Site (2021)
• Innovation and Entrepreneurship Pacesetter (2021-2022)
• Dean’s List (ZJUI) (2021-2022)
• Dean’s List (UIUC) (2023-2024)

PUBLICATIONS

• Sirui Xu, Dongting Li, Yucheng Zhang, et al., “InterAct: Advancing Large-Scale Versatile 3D Human-Object Interaction Generation”, CVPR (Under Review)
• Enxin Song, Wenhao Chai, Guanhong Wang, Yucheng Zhang, et al., “MovieChat: From Dense Token to Sparse Memory for Long Video Understanding”, CVPR 2024

RESEARCH EXPERIENCE

Human-Object Interaction with Skinned Multi-Person Linear Models | Research Assistant

Feb. 2024 - Present
Advisor: Yuxiong Wang (Professor, UIUC)

• Proposed a new method for representing human poses with a small number of markers.
• Built a data collection website on Amazon Mechanical Turk to create a dataset with 22.94 hours of human-object interaction data and detailed annotations.
• Processed data by segmenting video based on motion changes using Posecode and GPT-driven annotation.
• Performed penetration evaluation and motion prediction on the dataset.

Long Video Understanding | Team Leader

Jul. 2023 - Dec. 2023
Advisor: Gaoang Wang (Professor, ZJU-UIUC Institute)

• Developed a new memory management method to solve long video understanding problems with high computational load and memory cost.
• Reduced memory costs by selectively storing short-term memory and long-term memory based on frame similarity.
• Built a dataset of 1,000 long videos and manually annotated them with summaries and Q&A.
• Paper “MovieChat: From Dense Token to Sparse Memory for Long Video Understanding” accepted by CVPR 2024.

Temporal Action Detection under Weak Supervision | Team Leader

Mar. 2023 - Oct. 2023
Advisor: Gaoang Wang (Professor, ZJU-UIUC Institute)

• Modified the model using techniques from language models to reduce the accuracy gap between supervised and unsupervised models.
• Used attention mechanisms and transformers to detect and identify motions using weakly labeled data.

DNN Encryption | Individual Research

Jul. 2021 - Aug. 2021
Advisor: Jian Liu (Professor, Zhejiang University)

• Modified a two-party deep neural network inference model using lattice-based homomorphic encryption.
• Improved performance by utilizing vector-oblivious linear evaluation (VOLE)-styled oblivious transfer (OT) to optimize data transfer in ResNet-50.

PROJECTS

Simplified Computer System

• Developed a simplified computer system in C and assembly language that supports parallel execution, instruction input, storage, and program calls.
• Implemented drivers, RTC, system calls, file system, scheduling, memory mapping, and signals.

Traffic Light Simulation Control System

• Utilized YOLO to detect vehicles at intersections from video feeds and designed an algorithm to dynamically adjust traffic light duration based on vehicle presence and waiting time.
• Built a server to run YOLO and light control logic on FPGA, modified light duration accordingly, and visualized results with LED indicators.

TEACHING EXPERIENCE

MATH 241 Calculus III | Teaching Assistant

Fall 2024

• Led weekly discussion sections, office hours, and review sessions.
• Graded homework and exams.

SKILLS AND TOOLS

• Programming: MATLAB, C/C++, Python, Sagemath
• Frameworks & Tools: Pandas, PyTorch, smplx, Keras, NumPy, SciPy, tqdm