Zijian Chen
PhD Student at Shanghai Jiao Tong University
I am currently a 3rd PhD Student of Multimedia Lab at Shanghai Jiao Tong University (SJTU), advised by Prof. Guangtao Zhai and Prof. Wenjun Zhang. Previously, I received my M.E. degree from East China University of Science and Technology in June 2023, and B.S. degree in EE from Wenzhou University in June 2020. I am grateful to all the teachers and mentors who have supported me along the way and inspired my passion for research.
To understand what I'm thinking about right now, read in order:
- MLLMs in Unified Perception and Generation.
- Toward human–model alignment in fundamental capabilities.
- The perceptual limits/redundancies of MLLMs.
- Quality Assessment of Generative Videos.
- Deciphering oracle bone inscriptions: OBI-Bench and survey.
Some other notable things worth mentioning:
- I have made a lot of benchmarks: FigCodeBench, OBI-Bench, GAIA, etc.
- I previously worked as an intern for model evaluation in the OpenCompass team at the Shanghai Artificial Intelligence Laboratory for one and a half years.
- In 2025, I was selected for the Doctoral Student Program of the Young S&T Talents Cultivation Project, CAST.
I'm always eager to communicate and cooperate, so feel free to reach out at zijian.chen@sjtu.edu.cn.
news
| 2026.10 | Finished my internship at Shanghai Artificial Intelligence Laboratory, working on large model evaluation. |
|---|---|
| 2026.10 | FigCodeBench, our new benchmark for evaluating MLLMs on figure generation, is released! |
| 2026.09 | Two paper have been accept by NeurIPS 2026. |
| 2026.09 | One paper has been accept by EMNLP 2026. |
| 2026.09 | Our ROOTS has been accepted by npj Heritage Science |
latest posts
| Jun 11, 2025 | Gaining Inspiration from Darwinian Evolution: Towards Self-Referential and Self-Improving AI System |
|---|
selected publications
arxiv, 2026 (NEW)We propose FigCodeBench, a comprehensive framework for rigorously evaluating MLLMs on figure reproduction, integrating multimodal comprehension and generation.
npj Heritage Science, 2026 (NEW)We conduct a comprehensive survey into Oracle bone inscriptions information processing works over the past 20 years, reviewing more than 150 related articles.
ICASSP, 2026 (NEW)In this paper, we introduce PictOBI-20k, a dataset designed to evaluate LMMs on the visual decipherment tasks of pictographic OBCs. It includes 20k meticulously collected OBC and real object images, forming over 15k multi-choice questions. We also conduct subjective annotations to investigate the consistency of the reference point between humans and LMMs in visual reasoning.
arXiv, 2025In this paper, we introduce SQUID GAME, a dynamic and adversarial evaluation environment with resource-constrained and asymmetric information settings elaborated to evaluate LLMs through interactive gameplay against other LLM opponents.
arXiv, 2025In this paper, we introduce MACEval, a Multi-Agent Continual Evaluation network for dynamic evaluation of large models, and define a AUC-inspired metric to quantify performance longitudinally and sustainably.
arXiv, 2025In this paper, we introduce BioMotion Arena, the first biological motion-based visual preference evaluation framework for large models. We focus on ten typical human motions and introduce fine-grained control over gender, weight, mood, and direction. More than 45k votes for 53 mainstream LLMs and MLLMs on 90 biological motion variants are collected.
arXiv, 2025In this paper, we propose a novel concept, LMM-JND, to quantify the perceptual redundancy characteristic for LMMs and a well-designed pipeline for its determination. We also construct a large-scale dataset, named VPA-JND, which contains 21.5k reference images with over 489k stimuli across 12 distortion types, to facilitate LMM-JND studies.
IEEE Transactions on Circuits and Systems for Video Technology, 2025In this paper, we propose a unified deep neural network that explicitly disentangles the luminance and chrominance channels, and simultaneously recovers intensity gradients and color discontinuity from detection-free measurement in an end-to-end manner.
IEEE Transactions on Circuits and Systems for Video Technology, 2025In this work, we construct the AI-Generated Image Naturalness (AGIN) dataset and propose the Joint Objective Image Naturalness evaluaTor (JOINT) to automatically assess the naturalness of AIGIs that align with human opinions.
ICLR, 2025In this work, we introduce OBI-Bench, a holistic benchmark crafted to systematically evaluate large multi-modal models (LMMs) on whole-process oracle bone inscriptions (OBI) processing tasks demanding expert-level domain knowledge and deliberate cognition.
NeurIPS, 2024 (Spotlight Presentation)In this work, we construct GAIA, a Generic AI-generated Action dataset, by conducting a large-scale subjective evaluation from a novel causal reasoning-based perspective, resulting in 971,244 ratings among 9,180 video-action pairs, and evaluate a suite of popular text-to-video models on their ability to generate visually rational actions.
IEEE Transactions on Circuits and Systems for Video Technology, 2024In this work, we build the Banding Artifact Noticeable Database (BAND-2k), which consists of 2,000 banding images generated by 15 compression and quantization schemes.
IEEE International Symposium on Circuits and Systems (ISCAS), 2024In this paper, we develop a no-reference banding evaluator for banding detection and quality assessment by leveraging its frequency characteristics.