Introduction

Hi there! I’m Wouter, a Research Scientist at Google DeepMind in London working on Gemini. My research interests are multimodal learning, reinforcement learning (RL), and generative learning.

I obtained my Master of Science in Engineering from KU Leuven (Belgium). Shortly afterwards, I pursued a PhD degree in Computer Vision at KU Leuven under the supervision of Professor Luc Van Gool. In 2023, I successfully defended my PhD, which investigated self-supervised learning for visual scene understanding. During this time, I had the opportunity to intern at Meta (Facebook AI Research) in the San Francisco Bay Area.

In 2024, I started a Research Scientist position at Google DeepMind in London. At Google, I aim to improve Gemini’s visual perception and reasoning capabilities.

In my spare time, I enjoy hiking and cycling. Don’t hesitate to reach out!

News

• Dec. 2024: We released an experimental Gemini Flash Thinking model with improved reasoning capabilities. Try it out in Google AI Studio.
• April 2024: In April, I will be joining Google DeepMind in London.
• March 2024: New paper on panoptic segmentation with latent diffusion models: LDMSeg, accepted at ECCV.
• May 2023: Outstanding Reviewer for CVPR 2023.
• Oct. 2022: Outstanding Reviewer for ECCV 2022.
• Febr. 2022: Will be joining Facebook AI Research (Accel) in Menlo Park as a Research Intern during the summer.
• Sept. 2021: 1 paper accepted at NeurIPS.
• July 2021: 1 paper accepted at ICCV.
• June 2021: New paper on representation learning is released. Code is available on Github.
• April 2021: We organize a workshop on multi-task learning as a full-day event at ICCV.
• Febr. 2021: New paper on self-supervised semantic segmentation is released. Code is available on Github.
• Jan. 2021: Survey on multi-task learning is accepted at TPAMI.
• July 2020: Code and pretrained models are released for SCAN. It achieves >20% absolute improvement over previous works and even outperforms several semi-supervised methods.
• July 2020: 1 paper accepted at ECCV.
• July 2020: 1 paper accepted at RA-L + IROS.
• June 2020: Our paper “SCAN: Learning to Classify Images without Labels” is getting some attention on YouTube and Reddit.
• July 2019: Enjoyed the interesting talks at ICVSS 2019. I highly recommend it.
• May 2019: Our method reaches first place on the depth completion benchmark of KITTI. Code and pretrained model are released.

Research

Publications

Wouter Van Gansbeke, Bert De Brabandere. “A Simple Latent Diffusion Approach for Panoptic Segmentation and Mask Inpainting.” European Conference on Computer Vision (ECCV), 2024.
[ArXiv] [Code] [pdf]

Wouter Van Gansbeke, Simon Vandenhende, Luc Van Gool. “Discovering Object Masks with Transformers for Unsupervised Semantic Segmentation.” Arxiv Preprint, 2022.
[ArXiv] [Code] [pdf]

Wouter Van Gansbeke, Simon Vandenhende, Stamatios Georgoulis, Luc Van Gool. “Revisiting Contrastive Methods for Unsupervised Learning of Visual Representations.” Advances in Neural Information Processing Systems (NeurIPS), 2021.
[ArXiv] [Code] [pdf]

Wouter Van Gansbeke, Simon Vandenhende, Stamatios Georgoulis, Luc Van Gool. “Unsupervised Semantic Segmentation by Contrasting Mask Proposals.” International Conference on Computer Vision (ICCV), 2021.
[ArXiv] [Code] [pdf]

Simon Vandenhende, Stamatios Georgoulis, Wouter Van Gansbeke, Marc Proesmans, Dengxin Dai, Luc Van Gool. “Multi-Task Learning for Dense Prediction Tasks: A Survey.” IEEE Transactions on Pattern Analysis and Machine Intelligence (T-PAMI), 2020.
[ArXiv] [Code] [pdf]

Wouter Van Gansbeke, Simon Vandenhende, Stamatios Georgoulis, Marc Proesmans, Luc Van Gool. “SCAN: Learning to Classify Images without Labels.” European Conference on Computer Vision (ECCV), 2020.
[ArXiv] [Code] [pdf] [Video] [slides]

@article{wvangansbeke2020_learning,
  title={SCAN: Learning to Classify Images without Labels},
  author={Van Gansbeke, Wouter and Vandenhende, Simon and Georgoulis, Stamatios and Proesmans, Marc and Van Gool, Luc},
  journal={arXiv preprint arXiv:2005.12320},
  year={2020}
}

Wouter Van Gansbeke, Davy Neven, Bert De Brabandere, Luc Van Gool. “Sparse and Noisy LiDAR Completion with RGB Guidance and Uncertainty.” International Conference on Machine Vision Applications (MVA), 2019.
[ArXiv] [Code] [pdf] [Video]

@inproceedings{wvangansbeke_depth_2019,
    author={Van Gansbeke, Wouter and Neven, Davy and De Brabandere, Bert and Van Gool, Luc},
    booktitle={2019 16th International Conference on Machine Vision Applications (MVA)},
    title={Sparse and Noisy LiDAR Completion with RGB Guidance and Uncertainty},
    year={2019},
    pages={1-6},
    organization={IEEE}
}

Vaishakh Patil, Wouter Van Gansbeke, Dengxin Dai, Luc Van Gool “Don’t Forget The Past: Recurrent Depth Estimation from Monocular Video.” IEEE Robotics and Automation Letters, 2020.
[ArXiv] [pdf]

@article{patil2020_recurrent,
  title={Don’t Forget The Past: Recurrent Depth Estimation from Monocular Video},
  author={Vaishakh, Patil and Van Gansbeke, Wouter and Dai, Dengxin and Van Gool, Luc},
  booktitle={arXiv preprint arXiv:2001.02613},
  year={2020}
}

Wouter Van Gansbeke, Bert De Brabandere, Davy Neven, Marc Proesmans, Luc Van Gool. “End-to-end Lane Detection through Differentiable Least-Squares Fitting.” International Conference on Computer Vision Workshops (ICCV Workshops), 2019.
[ArXiv] [Code] [pdf]

@inproceedings{wvangansbeke2019_end,
  title={End-to-end Lane Detection through Differentiable Least-Squares Fitting},
  author={Van Gansbeke, Wouter and De Brabandere, Bert and Neven, Davy and Proesmans, Marc and Van Gool, Luc},
  booktitle={Proceedings of the IEEE International Conference on Computer Vision (ICCV) Workshops},
  year={2019}
}