About Me
Zeyuan Zang臧泽元
B.Eng. Student @ BUPT • Incoming Ph.D. @ ZGCA
I am a 4th-year undergraduate student majoring in Computer Science and Technology at the School of Future, Beijing University of Posts and Telecommunications (BUPT), currently conducting research on Large Vision-Language Models (LVLMs) at the Center of Intelligence Science and Technology (CIST), advised by Prof. Xiaojie Wang. I will soon begin my Ph.D. through a joint program between BUPT and the Zhongguancun Academy (ZGCA), advised by Prof. Wentao Zhang.
My current research interests include:
- Boosting reasoning abilities of Large Vision-Language Models (LVLMs)
- Vision-Language representation learning
- Novel architectures inspired by cognitive heuristics
- Improving LVLMs' reliability with high-quality data
Outside the lab, I am a classical music enthusiast and enjoy tennis and golf.
News
- 2026.01 Joined Zhongguancun Academy as a research intern, working on multimodal post-training data synthesis for physics.
- 2025.09 Admitted to the direct Ph.D. program at Zhongguancun Academy.
- 2025.09 Reading Images Like Texts accepted to ICLR 2026.
- 2025.07 VL-DynaRefine accepted to ACM MM 2025; HADAR on display at WSIS+20 & AI for Good Summit, Geneva.
Research & Publications
Research Grants
Topic: Research and Application of Event Causality Identification Based on Causal Graphs
- Responsible for implementing and evaluating a causal graph-based causal reasoning model.
Topic: Optimization of Vision-Language Models for Fine-Grained Perception Tasks
- Organized and executed the project, leading method design, implementation, and evaluation.
Research Experience
- Building a multimodal post-training data synthesis pipeline for the physics domain.
- Fine-grained visual perception of LVLMs via multi-crop in-context learning.
- Visual reasoning with real-time verification and refinement of local sub-tasks.
- Mitigating object hallucination by re-aligning decoder hidden states with the visual context.
- Linguistic steganography and steganalysis in the deep learning era: feature extraction via word-order shuffling and token-level semantic comparison.
- Studied the processing and analysis of atomic-resolution images captured by scanning transmission electron microscopy (STEM).
- Gesture recognition from vision and somatosensory data; BCI- and VR-based rehabilitation robotics.
Selected Projects
- Deepseek-OCR implemented in vLLM backend for document parsing, with tweaks (concurrency control, Lossless PDF with bounding box, markdown with clearer images) added.
- Planning to integrate with LangChain to build an agentic document reading system.
- Developed a lightweight object dropping detection algorithm achieving 6-10 FPS real-time performance and 75% accuracy on Raspberry Pi.
- Designed and implemented an incident recording system for surveillance applications, composed of a responsive H5 frontend built with Vue.js and a scalable backend service built with Node.js.
A multimodal digital human interaction frontend application based on React + TypeScript + WebRTC, supporting speech, gesture, facial expression recognition, and real-time conversation features.
Education
- GPA: 3.79/4 (Score: 90.92/100, Rank: 5/30) | IELTS: 7.5, CET-6: 615, CET-4: 669
- Honors: 1st Class Scholarship (2025), Xiaomi Scholarship (1st class, 2024), Merit Student of Beijing (2024), Merit Student of BUPT (2024, 2025)
Graduated at 16.
Tech Stack
