|
Chenyi Zhuang INTJ / 二次元 / 咖啡和猫 I'm a third-year master's student in Computer Technology at the Nanjing University of Aeronautics and Astronautics (2022-2025), advised by Prof. Pan Gao at Immersive and Interactive Multimedia Lab (I2ML). My current research interests are image-related (multi-modal) vision tasks, especially generative models and their explainability. I am also interested in computer graphics and video understanding. NEWS! I am applying for PhD in Fall 2025, feel free to contact me ⬇ WeChat / Email / CV / Google Scholar / Github |
|
Research* indicate an equal contribution. |
|
DiffuseST: Unleashing the Capability of the Diffusion Model for Style Transfer
Ying Hu*, Chenyi Zhuang*, Pan Gao. ACM MM Asia 2024 Leverage textual and spatial representations and the step-by-step denoising nature of the pre-trained diffusion model to achieve balanced style transfer results. code / arXiv |
|
Magnet: We Never Know How Text-to-Image Diffusion Models Work, Until We Learn How Vision-Language Models Function
Chenyi Zhuang, Ying Hu, Pan Gao. NeurIPS 2024 In-depth analysis of attribute understanding for CLIP text encoder and CLIP-based diffusion models, a novel training-free approach to tackle the attribute binding issue. code / arXiv |
|
CDFormer: When Degradation Prediction Embraces Diffusion Model for Blind Image Super-Resolution
Qingguo Liu, Chenyi Zhuang, Pan Gao, Jie Qin. CVPR 2024 Investigate the diffusion model as an estimator to predict Content Degradation Prior (CDP) with rich content detail for the super-resolution task. code / arXiv |
|
StylePrompter: All Styles Need Is Attention
Chenyi Zhuang, Pan Gao, Aljosa Smolic. ACM MM 2023 Propose a Transformer-based framework to predict W+ codes at the token level for StyleGAN, with a Style-driven Multi-scale Adaptive Refinement Transformer (SMART) block to refine features in F space. code / arXiv |
What's More?My academic career is guided by two principles: "A believing heart is a magic" as my faith and commitment to any upcoming event; and "simple but effective" as a guideline for my work style with a high level of productivity and creativity. |