Advancing
Visual Intelligence
for a Better World.
We are a research lab at The Hong Kong Polytechnic University led by Chair Professor Lei Zhang, working on the frontiers of computer vision, multimodal foundation models, and generative AI.
“Y learning and beyond — for future visual enhancement and understanding.” — VCLab Mission
Rooted at The Hong Kong Polytechnic University.
Our lab sits inside the Department of Computing on PolyU's brick‑red Hung Hom campus — home to Li Ka Shing Tower, the Jockey Club Innovation Tower by Zaha Hadid, and a vibrant community of researchers pushing the boundaries of science and design.
A home for curious visual-intelligence researchers.
Founded at the Department of Computing, PolyU, VCLab explores both fundamental and applied problems in visual computing, with strong industrial collaboration and a global publication footprint.
The Visual Computing Lab (VCLab) at The Hong Kong Polytechnic University pursues research across the full stack of modern visual intelligence — from low-level image and video restoration, through multimodal perception and reasoning, to large-scale generative models, 3D reconstruction, new network architectures and rigorous evaluation benchmarks.
Our work is regularly published in top-tier venues including CVPR, ICCV, ECCV, NeurIPS, ICLR, TPAMI and IJCV, and many of our methods have been deployed on hundreds of millions of mobile devices through our long-term collaboration with OPPO Research Institute.
We welcome talented students, postdocs and interns who care deeply about visual intelligence. If our mission resonates with you, come join us →
Six research directions.
One unified vision.
Our research is organized into six complementary collections. Click any card to see related publications, code and demos.
Image / Video Restoration, Enhancement & Quality Assessment
Leading research on real-world super-resolution, denoising, deblurring, HDR, flicker removal and perceptual quality assessment.
Explore collection → 02Multimodal Perception, Understanding & Reasoning
MLLM-driven visual perception, grounding and reasoning — enhancing the multi-tasking capabilities of large multimodal models.
Explore collection → 03Image & Video Synthesis and Generation
Accelerating, distilling and improving diffusion / autoregressive / DiT-based generative models for images and videos.
Explore collection → 043D Perception, Reconstruction & Generation
Sensing, reconstructing, synthesizing and editing high-fidelity 3D worlds from images, videos and language prompts.
Explore collection → 05Architecture & Training Paradigms
Novel architectures of vision transformers, LLMs / VLMs and efficient, decentralized training paradigms for frontier models.
Explore collection → 06Benchmarks & Datasets
Evaluation benchmarks and training datasets driving rigorous, reproducible progress for the visual computing community.
Explore collection →What's happening at VCLab.
Recent work from our lab.
A curated selection of recent VCLab publications. For the full list, see Prof. Zhang's publication page ↗.
People behind the work.
A close-knit team of researchers, PhDs and interns passionate about pushing visual intelligence forward.
Our alumni are now at leading universities and companies including MSRA, Huawei Noah's Ark Lab, Meituan, Tencent AI Lab, Alibaba DAMO, OPPO, ByteDance and more.
Join us in shaping
the future of visual intelligence.
We are recruiting PhD students (jointly trained with OPPO Research Institute), postdocs and research interns across all six directions — image / video restoration, generative models, multimodal reasoning, 3D, efficient architectures and benchmarks.
Get in touch.
Prof. Lei Zhang
Department of Computing
The Hong Kong Polytechnic University
Hung Hom, Kowloon, Hong Kong
Office: PQ816
Email: cslzhang [at] comp.polyu.edu.hk
For general lab inquiries, collaborations and press, please email Prof. Lei Zhang. For prospective students, please include your CV and a brief statement of interest.