publications | Tao Hu

2026

Guiding Token-Sparse Diffusion Models

Felix Krause, Stefan Andreas Baumann, Johannes Schusterbauer, and 4 more authors

2026

@article{krause2026guiding,
  title = {Guiding Token-Sparse Diffusion Models},
  author = {Krause, Felix and Baumann, Stefan Andreas and Schusterbauer, Johannes and Grebenkova, Olga and Gui, Ming and Hu, Vincent Tao and Ommer, Björn},
  year = {2026},
  booktitle = {Arxiv},
}

Purrception: Variational Flow Matching for Vector-Quantized Image Generation

Răzvan-Andrei Matișan, Vincent Tao Hu, Grigory Bartosh, and 6 more authors

In ICLR, 2026

Bib PDF

@inproceedings{maticsan2026purrception,
  title = {Purrception: Variational Flow Matching for Vector-Quantized Image Generation},
  author = {Matișan, Răzvan-Andrei and Hu, Vincent Tao and Bartosh, Grigory and Ommer, Björn and Snoek, Cees G.M. and Welling, Max and van de Meent, Jan-Willem and Derakhshani, Mohammad Mahdi and Eijkelboom, Floor},
  year = {2026},
  booktitle = {ICLR},
}

Diffusion Models and Representation Learning: A Survey

Michael Fuest, Pingchuan Ma, Ming Gui, and 3 more authors

In T-PAMI, 2026

The interplay between diffusion models and representation learning

Bib PDF Code

@inproceedings{diffusion_rl_survey,
  title = {Diffusion Models and Representation Learning: A Survey},
  author = {Fuest, Michael and Ma, Pingchuan and Gui, Ming and Fischer, Johannes and Hu, Vincent Tao and Ommer, Bjorn},
  year = {2026},
  note = {The interplay between diffusion models and representation learning},
  booktitle = {T-PAMI},
  repostar = {dongzhuoyao/Diffusion-Representation-Learning-Survey-Taxonomy}
}

2025

TREAD: Token Routing for Efficient Architecture-agnostic Diffusion Training

Felix Krause, Timy Phan, Ming Gui, and 3 more authors

In ICCV, 2025

Bib PDF Code Website

@inproceedings{tread,
  title = {TREAD:  Token Routing for Efficient Architecture-agnostic Diffusion Training},
  author = {Krause, Felix and Phan, Timy and Gui, Ming and Baumann, Stefan Andreas and Hu, Vincent Tao and Ommer, Björn},
  year = {2025},
  note11 = {Skipping layers to boost training convergence},
  booktitle = {ICCV},
  highlight111 = {Oral},
  repostar = {CompVis/tread}
}

Stochastic Interpolants for Revealing Stylistic Flows across the History of Art

Pingchuan Ma, Ming Gui, Johannes Schusterbauer, and 4 more authors

In ICCV, 2025

Bib PDF Code Website

@inproceedings{artflow,
  title = {Stochastic Interpolants for Revealing Stylistic Flows across the History of Art},
  author = {Ma, Pingchuan and Gui, Ming and Schusterbauer, Johannes and Yang, Xiaopei and Grebenkova, Olga and Hu, Vincent Tao and Ommer, Björn},
  year = {2025},
  booktitle = {ICCV},
  repostar = {CompVis/Art-fm}
}

Continuous, Subject-Specific Attribute Control in T2I Models by Identifying Semantic Directions

Stefan Andreas Baumann, Felix Krause, Michael Neumayr, and 3 more authors

In CVPR, 2025

Prompt Editing in T2I models

Bib PDF Code Website

@inproceedings{baumann2024attributecontrol,
  title = {{C}ontinuous, {S}ubject-{S}pecific {A}ttribute {C}ontrol in {T}2{I} {M}odels by {I}dentifying {S}emantic {D}irections},
  author = {Baumann, Stefan Andreas and Krause, Felix and Neumayr, Michael and Stracke, Nick and Hu, Vincent Tao and Ommer, Björn},
  year = {2025},
  note = {Prompt Editing in T2I models},
  booktitle = {CVPR},
  repostar = {CompVis/attribute-control}
}

MaskFlow: Discrete Flows for Flexible and Efficient Long Video Generation

Michael Fuest, Vincent Tao Hu, and Björn Ommer

In Arxiv, 2025

Bib PDF Code Website

@inproceedings{maskflow,
  title = {MaskFlow: Discrete Flows for Flexible and Efficient Long Video Generation},
  author = {Fuest, Michael and Hu, Vincent Tao and Ommer, Björn},
  year = {2025},
  booktitle = {Arxiv},
  preview1111 = {./mask.jpg},
  repostar = {CompVis/maskflow}
}

ToddlerDiffusion: Flash Interpretable Controllable Diffusion Model

Eslam Mohamed BAKR, Liangbing Zhao, Vincent Tao Hu, and 3 more authors

In ICLR, 2025

Bib PDF Website

@inproceedings{ToddlerDiffusion,
  title = {ToddlerDiffusion: Flash Interpretable Controllable Diffusion Model},
  author = {BAKR, Eslam Mohamed and Zhao, Liangbing and Hu, Vincent Tao and Cord, Matthieu and Perez, Patrick and Elhoseiny, Mohamed},
  year = {2025},
  booktitle = {ICLR},
  repostar = {toddlerdiffusion/code}
}

DepthFM: Fast Monocular Depth Estimation with Flow Matching

Ming Gui, Johannes S. Fischer, Ulrich Prestel, and 6 more authors

In AAAI, 2025

An exploration of flow matching for blazing fast and zero-shot depth estimation

Bib PDF Code Website

@inproceedings{depthfm,
  title = {DepthFM: Fast Monocular Depth Estimation with Flow Matching},
  author = {Gui, Ming and Fischer, Johannes S. and Prestel, Ulrich and Ma, Pingchuan and Kotovenko, Dmytro and Grebenkova, Olga and Baumann, Stefan A. and Hu, Vincent Tao and Ommer, Björn},
  year = {2025},
  note = {An exploration of flow matching for blazing fast and zero-shot depth estimation},
  booktitle = {AAAI},
  highlight = {Oral},
  repostar = {CompVis/depth-fm}
}

Does VLM Classification Benefit from LLM Description Semantics?

Pingchuan Ma, Lennart Rietdorf, Dmytro Kotovenko, and 2 more authors

In AAAI, 2025

PDF Code Website

Distillation of Diffusion Features for Semantic Correspondence

Frank Fundel, Johannes Schusterbauer, Vincent Tao Hu, and 1 more author

In WACV, 2025

Bib PDF Code Website

@inproceedings{fundel2025distilldift,
  title = {Distillation of Diffusion Features for Semantic Correspondence},
  author = {Fundel, Frank and Schusterbauer, Johannes and Hu, Vincent Tao and Ommer, Björn},
  year = {2025},
  note = {},
  booktitle = {WACV},
  repostar = {CompVis/DistillDIFT}
}

2024

[MASK] is All You Need

Vincent Tao Hu and Björn Ommer

In Arxiv, 2024

Bib PDF Code Website

Scaling Image Tokenizers with Grouped Spherical Quantization

Jiangtao Wang, Zhen Qin, Yifan Zhang, and 4 more authors

2024

Bib PDF Code

@article{wang2024scaling,
  title = {Scaling Image Tokenizers with Grouped Spherical Quantization},
  author = {Wang, Jiangtao and Qin, Zhen and Zhang, Yifan and Hu, Vincent Tao and Ommer, Bj{\"o}rn and Briq, Rania and Kesselheim, Stefan},
  booktitle = {Arxiv},
  year = {2024},
  note = {},
  repostar = {HelmholtzAI-FZJ/flex_gen}
}

ZigMa: A DiT-style Zigzag Mamba Diffusion Model

Vincent Tao Hu, Stefan Andreas Baumann, Ming Gui, and 4 more authors

In ECCV, 2024

a DiT-style Mamba-based diffusion models

Bib PDF Code Website

@inproceedings{hu2024zigmaa,
  title = {ZigMa: A DiT-style Zigzag Mamba Diffusion Model},
  author = {Hu, Vincent Tao and Baumann, Stefan Andreas and Gui, Ming and Grebenkova, Olga and Ma, Pingchuan and Fischer, Johannes and Ommer, Bjorn},
  year = {2024},
  note = {a DiT-style Mamba-based diffusion models },
  booktitle = {ECCV},
  repostar = {CompVis/zigma}
}

Boosting Latent Diffusion with Flow Matching

Johannes S. Fischer, Ming Gui, Pingchuan Ma, and 4 more authors

In ECCV, 2024

flow matching for super-resolution

Bib PDF Code Website

@inproceedings{fischer2023boosting,
  title = {Boosting Latent Diffusion with Flow Matching},
  author = {Fischer, Johannes S. and Gui, Ming and Ma, Pingchuan and Stracke, Nick and Baumann, Stefan A. and Hu, Vincent Tao and Ommer, Bjorn},
  year = {2024},
  note = {flow matching for super-resolution},
  booktitle = {ECCV},
  highlight = {Oral},
  repostar = {CompVis/fm-boosting}
}

Guided Flow Vision Transformer from Self-Supervised Diffusion Features

Vincent Tao Hu, Yunlu Chen, Mathilde Caron, and 3 more authors

In Arxiv, 2024

Bib PDF Code Website

@inproceedings{sgfm,
  title = {Guided Flow Vision Transformer from Self-Supervised Diffusion Features},
  author = {Hu, Vincent Tao and Chen, Yunlu and Caron, Mathilde and Asano, Yuki M. and Snoek, Cees G.M. and Ommer, Björn},
  year = {2024},
  note = {},
  booktitle = {Arxiv},
  repostar = {dongzhuoyao/sgfm}
}

Motion Flow Matching for Human Motion Synthesis and Editing

Vincent Tao Hu, Wenzhe Yin, Pingchuan Ma, and 7 more authors

In Arxiv, 2024

Bib PDF Code Website

@inproceedings{motionfm,
  title = {Motion Flow Matching for Human Motion Synthesis and Editing},
  author = {Hu, Vincent Tao and Yin, Wenzhe and Ma, Pingchuan and Chen, Yunlu and Fernando, Basura and Asano, Yuki M. and Gavves, Efstratios and Mettes, Pascal and Ommer, Björn and Snoek, Cees G.M.},
  year = {2024},
  note1 = {},
  booktitle = {Arxiv},
  repostar = {dongzhuoyao/motionfm}
}

Training Class-Imbalanced Diffusion Model Via Overlap Optimization

Divin Yan, Lu Qi, Vincent Tao Hu, and 2 more authors

In arxiv, 2024

Bib PDF

@inproceedings{constrastivediffusion,
  title = {Training Class-Imbalanced Diffusion Model Via Overlap Optimization},
  author = {Yan, Divin and Qi, Lu and Hu, Vincent Tao and Yang, Ming-Hsuan and Tang, Meng},
  year = {2024},
  note = {},
  booktitle = {arxiv},
}

Latent Space Editing in Transformer-based Flow Matching

Vincent Tao Hu, David W Zhang, Pascal Mettes, and 3 more authors

In AAAI 2024. Also appear in ICML 2023 Workshop, New Frontiers in Learning, Control, and Dynamical Systems, 2024

Bib PDF Code Poster Website

@inproceedings{hulfm,
  title = {Latent Space Editing in Transformer-based Flow Matching},
  author = {Hu, Vincent Tao and Zhang, David W and Mettes, Pascal and Tang, Meng and Zhao, Deli and Snoek, Cees G.M.},
  year = {2024},
  note = {},
  booktitle = {AAAI 2024. Also appear in ICML 2023 Workshop, New Frontiers in Learning, Control, and Dynamical Systems},
  repostar = {dongzhuoyao/uspace}
}

Flow Matching for Conditional Text Generation in a Few Sampling Steps

Vincent Tao Hu, Di Wu, Yuki M. Asano, and 4 more authors

In EACL, 2024

Flow Matching for text generation

Bib PDF Code Website

@inproceedings{huflowseq,
  title = {Flow Matching for Conditional Text Generation in a Few Sampling Steps},
  author = {Hu, Vincent Tao and Wu, Di and Asano, Yuki M. and Mettes, Pascal and Fernando, Basura and Ommer, Björn and Snoek, Cees G.M.},
  year = {2024},
  booktitle = {EACL},
  note = {Flow Matching for text generation},
  selectedddd = {true},
  repostar = {dongzhuoyao/flowseq}
}

2023

Self-Guided Diffusion Models

Tao Hu^*, David W Zhang^*, Yuki M. Asano, and 2 more authors

In CVPR, 2023

A bridge between the community of self-supervised learning and diffusion models. Short version to appear in NeurIPS 2022 Workshop on Score-Based Methods and NeurIPS 2022 Workshop Self-Supervised Learning Theory and Practice.

Bib PDF Code Website

@inproceedings{hu2022selfguided,
  title = {Self-Guided Diffusion Models},
  author = {Hu*, Tao and Zhang*, David W and Asano, Yuki M. and Burghouts, Gertjan J. and Snoek, Cees G.M.},
  year = {2023},
  booktitle = {CVPR},
  note = {A bridge between the community of self-supervised learning and diffusion models. Short version to appear in NeurIPS 2022 Workshop on Score-Based Methods and
                   NeurIPS 2022 Workshop Self-Supervised Learning Theory and Practice.},
  repostar = {dongzhuoyao/self-guided-diffusion-models}
}

2021

Self-supervised Video Representation Learning with Cross-Stream Prototypical Contrasting

Martine Toering, Ioannis Gatopoulos, Maarten Stol, and 1 more author

In WACV, 2021

Improve video representation by constrasting Prototypical features.

Bib PDF Code

@inproceedings{toering2022self,
  title = {Self-supervised Video Representation Learning with Cross-Stream Prototypical Contrasting},
  author = {Toering, Martine and Gatopoulos, Ioannis and Stol, Maarten and Hu, Tao},
  booktitle = {WACV},
  year = {2021},
  note = {Improve video representation by constrasting Prototypical features.},
  repostar = {martinetoering/ViCC}
}

2020

Localizing the Common Action Among a Few Videos

Pengwan Yang^*, Tao Hu^*, Pascal Mettes, and 1 more author

In European Conference on Computer Vision(ECCV), 2020

Localizing the temporal extent of an action in a long untrimmed video by attention techniques.

Bib PDF Code

@inproceedings{YangECCV20,
  author = {Yang*, Pengwan and Hu*, Tao and Mettes, Pascal and Snoek, Cees G.M.},
  title = {Localizing the Common Action Among a Few Videos},
  booktitle = {European Conference on Computer Vision(ECCV)},
  year = {2020},
  note = {Localizing the temporal extent of an action in a long untrimmed video by attention techniques.},
  repostar = {PengWan-Yang/commonLocalization}
}

Pointmixup: Augmentation for point clouds

Yunlu Chen^*, Tao Hu^*, Efstratios Gavves, and 4 more authors

In European Conference on Computer Vision(ECCV), 2020

A simple augmentation method based on MixUp to boost the performance on related tasks of point cloud.

Bib PDF Code

@inproceedings{chen2020pointmixup,
  title = {Pointmixup: Augmentation for point clouds},
  author = {Chen*, Yunlu and Hu*, Tao and Gavves, Efstratios and Mensink, Thomas and Mettes, Pascal and Yang, Pengwan and Snoek, Cees G.M.},
  booktitle = {European Conference on Computer Vision(ECCV)},
  highlight = {Spotlight},
  year = {2020},
  note = {A simple augmentation method based on MixUp to boost the performance on related tasks of point cloud.},
  repostar = {yunlu-chen/PointMixup}
}

2019

Attention-based Multi-Context Guiding for Few-Shot Semantic Segmentation

Tao Hu, Pengwan Yang, Chiliang Zhang, and 3 more authors

In AAAI, 2019

Solve the few-shot segmentation problem by applying attention in multi-scales.

Bib PDF Code Poster

@inproceedings{tao2018fss,
  title = {Attention-based Multi-Context Guiding for Few-Shot Semantic Segmentation},
  author = {Hu, Tao and Yang, Pengwan and Zhang, Chiliang and Yu, Gang and Mu, Yadong and Snoek, Cees G.M.},
  booktitle = {AAAI},
  highlight = {Spotlight},
  year = {2019},
  note = {Solve the few-shot segmentation problem by applying attention in multi-scales.},
}

SILCO: Show a Few Images, Localize the Common Object

Tao Hu, Pascal Mettes, Jia-Hong Huang, and 1 more author

In International Conference on Computer Vision(ICCV), 2019

Design a graph network and apply attention on them to solve the problem of common object localization.

Bib PDF Code Poster Website

@inproceedings{silco2019,
  title = {SILCO: Show a Few Images, Localize the Common Object},
  author = {Hu, Tao and Mettes, Pascal and Huang, Jia-Hong and Snoek, Cees G.M.},
  booktitle = {International Conference on Computer Vision(ICCV)},
  year = {2019},
  note = {Design a graph network and apply attention on them to solve the problem of common object localization.},
  repostar = {dongzhuoyao/silco_}
}