PeopleofSuperintelligence

Researcher

Han Zhou

Han Zhou is a computer vision researcher whose work spans generative modeling, low-light image enhancement and 3D/4D scene reconstruction. At McMaster University, she developed non-homogeneous dehazing methods in 2023 and 2024, including DehazeDCT, which ranked second in the NTIRE 2024 Dense and Non-Homogeneous Dehazing Challenge. Her first-author papers GLARE, GPP-LLIE and LITA-GS appeared at ECCV 2024, AAAI 2025 and CVPR 2025. She was named a 2024–2025 RBC Borealis Fellow and received her PhD from McMaster University in November 2025. At the Vector Institute, she is a Distinguished Postdoctoral Fellow working with Leonid Sigal and Igor Gilitschenski.

Timeline

  1. 2023

    Non-homogeneous dehazing with ConvNeXt

    In 2023, she is the first author of "Breaking Through the Haze," an arXiv paper on non-homogeneous image dehazing based on fast Fourier convolution and ConvNeXt, coauthored with Wei Dong, Yangyi Liu and Jun Chen.

  2. 2024

    DehazeDCT in the NTIRE challenge

    With coauthors at McMaster University, she presents DehazeDCT in 2024, a non-homogeneous dehazing method using a deformable convolutional transformer. The method ranks second among 16 submissions in the NTIRE 2024 Dense and Non-Homogeneous Dehazing Challenge.

  3. 2024

    GLARE at ECCV 2024

    In 2024, she is the first-listed author of GLARE, a low-light image enhancement network that retrieves codes from a vector-quantized codebook built from undegraded normal-light images. The paper is accepted at ECCV 2024 and also tests the method as a preprocessing step for low-light object detection.

  4. 2024–2025

    An RBC Borealis Fellowship

    She is named a 2024–2025 RBC Borealis Fellow, one of ten PhD researchers selected from universities across Canada. Her fellowship research covers generative AI, 3D vision, multimodal AI and vision-language models, with Jun Chen as her faculty advisor.

  5. 2025

    GPP-LLIE at AAAI 2025

    The AAAI-25 paper "Low-Light Image Enhancement via Generative Perceptual Priors" appears in 2025 with Zhou as first-listed author at McMaster University, alongside Wei Dong, Xiaohong Liu, Yulun Zhang, Guangtao Zhai and Jun Chen. Its GPP-LLIE framework uses vision-language models to produce global and local perceptual priors that guide a diffusion-based enhancement model.

  6. 2025

    LITA-GS at CVPR 2025

    With Wei Dong and Jun Chen, she publishes LITA-GS at CVPR 2025, a method for novel view synthesis under adverse lighting that combines reference-free 3D Gaussian Splatting with illumination-invariant physical priors. It trains without ground-truth references and surpasses a NeRF-based method with faster inference.

  7. November 2025

    A PhD from McMaster

    She receives her PhD from McMaster University in November 2025. Her doctoral research focuses on computer vision, generative modeling, image enhancement, and 3D/4D scene reconstruction under challenging real-world conditions.

  8. As of October 2026

    Postdoc at the Vector Institute

    She is a Distinguished Postdoctoral Fellow at the Vector Institute, working with Leonid Sigal and Igor Gilitschenski. Her recent work includes illumination-robust novel view synthesis and feed-forward dynamic scene reconstruction, and her research interests include 3D/4D vision, world models and image and video restoration.

Career

  1. NowVector InstituteDistinguished Postdoctoral Fellow

Education

  1. 2025McMaster UniversityPhD

From the biography.

  1. Of 943 people shaping AI who worked at Google, 446 work there nowIncludes Vector Institute. Our estimates of where the 943 people shaping AI who worked at Google and Google DeepMind work now: 446 at Google and 452 across 278 other organizations.
  2. Nonprofit roles are 2.3 times as common in AI safetyIncludes Vector Institute. We estimate where people focused on safety, alignment and evaluation work. Nonprofit roles appear for 25.1% of safety people with known affiliations, compared with 11.0% in other areas.
  3. Universities (31%) and startups and other companies (25%) employ the largest shares of 2020–2026 entrantsIncludes Vector Institute. Among 819 people whose first recorded role started in 2020–2026 and who have a known current organization, 60% work in commercial sectors (25% at startups and other companies, 23% at frontier labs and 13% at big tech, with 9 in two), and 31% work at universities, compared with 66% for earlier entrants.
  4. The United States employs an estimated 48% of people shaping AIIncludes Vector Institute. We estimate where the people shaping AI work today, how universities, companies and research institutes split across countries, and how country shares have changed since the 1960s.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Han Zhou yet. Send us a free-licensed photo.

  1. [2305.04430] Breaking Through the Haze: An Advanced Non-Homogeneous Dehazing Method based on Fast Fourier Convolution and ConvNeXt arxiv.org. Checked 22 October 2026.
  2. [2407.05169] DehazeDCT: Towards Effective Non-Homogeneous Dehazing via Deformable Convolutional Transformer arxiv.org. Checked 22 October 2026.
  3. ECCV Poster GLARE: Low Light Image Enhancement via Generative Latent Feature based Codebook Retrieval eccv2024.ecva.net. Checked 22 October 2026.
  4. [2407.12431] GLARE: Low Light Image Enhancement via Generative Latent Feature based Codebook Retrieval arxiv.org. Checked 22 October 2026.
  5. The 2024-2025 RBC Borealis Fellows: Driving the Future of AI rbcborealis.com. Checked 22 October 2026.
  6. Han Zhou - Vector Institute for Artificial Intelligence vectorinstitute.ai. Checked 22 October 2026.
  7. Low-Light Image Enhancement via Generative Perceptual Priors ojs.aaai.org. Checked 22 October 2026.
  8. LITA-GS: Illumination-Agnostic Novel View Synthesis via Reference-Free 3D Gaussian Splatting and Physical Priors (CVPR 2025 Open Access) openaccess.thecvf.com. Checked 22 October 2026.
  9. [2504.00219] LITA-GS: Illumination-Agnostic Novel View Synthesis via Reference-Free 3D Gaussian Splatting and Physical Priors arxiv.org. Checked 22 October 2026.