Emad Bahrami

About

I’m a PhD candidate in the Computer Vision Group at the University of Bonn, advised by Prof. Jürgen Gall. My research focuses on multimodal AI and video understanding, with an emphasis on temporal reasoning, generative modeling, and efficient AI. My work spans multimodal LLMs, reinforcement learning, video generation, temporal action segmentation, and long-term video anticipation. I’m currently working on generative world models and visual-physical reasoning.

I’ve been a Research Intern at Microsoft Research, working on RL post-training of multimodal LLMs (GRPO) for video question answering and spatiotemporal reasoning, and at Toyota Motor Europe, working on video understanding for manufacturing applications.

Before my PhD, I was a researcher at Deep MI, working on semantic segmentation of human brain MRI, and a visiting researcher in Prof. Gall’s group, working on action recognition and future frame prediction. I received my bachelor’s degree from the University of Tehran.

News

Sep 2026 Three papers accepted: one at BMVC 2026 and two at CVIU.
Apr 2026 STRIVE, from my Microsoft Research internship, is on arXiv.
Nov 2025 I’ve joined Toyota Motor Europe as a Research Intern.
May 2025 Glad to be recognized as an Outstanding Reviewer at CVPR 2025 🎉
May 2025 I’ve joined Microsoft as a Research Intern.

Selected publications

All publications →

STRIVE: Structured Spatiotemporal Exploration for Reinforcement Learning in Video Question Answering

Emad Bahrami, Olga Zatsarynna, Parth Pathak, Sunando Sengupta, Juergen Gall, and Mohsen Fayyaz

arXiv 2026Under review

Work done during an internship at Microsoft Research.

Paper BibTeX

@article{bahrami2026strive,
  title = {STRIVE: Structured Spatiotemporal Exploration for Reinforcement Learning in Video Question Answering},
  author = {Bahrami, Emad and Zatsarynna, Olga and Pathak, Parth and Sengupta, Sunando and Gall, Juergen and Fayyaz, Mohsen},
  abbr = {arXiv},
  year = {2026},
  journal = {Under review},
  note = {Work done during an internship at Microsoft Research.},
  pdf = {https://arxiv.org/abs/2604.01824},
  selected = {true},
  preview = {STRIVE_Icon.png}
}

Towards Generalizing Temporal Action Segmentation to Unseen Views

Emad Bahrami*, Olga Zatsarynna*, Gianpiero Francesca, and Juergen Gall

IJCV 2025International Journal of Computer Vision (IJCV)* Equal contribution

Paper BibTeX

@article{bahrami2025unseen,
  title = {Towards Generalizing Temporal Action Segmentation to Unseen Views},
  author = {Bahrami*, Emad and Zatsarynna*, Olga and Francesca, Gianpiero and Gall, Juergen},
  abbr = {IJCV},
  year = {2025},
  journal = {International Journal of Computer Vision (IJCV)},
  pdf = {https://arxiv.org/abs/2504.02512},
  selected = {true},
  preview = {UnVAS.png}
}

MANTA: Diffusion Mamba for Efficient and Effective Stochastic Long-Term Dense Anticipation

Olga Zatsarynna, Emad Bahrami, Yazan Abu Farha, Gianpiero Francesca, and Juergen Gall

CVPR 2025IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

Paper BibTeX

@inproceedings{zatsarynna2025manta,
  title = {MANTA: Diffusion Mamba for Efficient and Effective Stochastic Long-Term Dense Anticipation},
  author = {Zatsarynna, Olga and Bahrami, Emad and Farha, Yazan Abu and Francesca, Gianpiero and Gall, Juergen},
  year = {2025},
  booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  pdf = {https://arxiv.org/abs/2501.08837},
  abbr = {CVPR},
  selected = {true},
  preview = {Manta.png}
}

Gated Temporal Diffusion for Stochastic Long-Term Dense Anticipation

Olga Zatsarynna*, Emad Bahrami*, Yazan Abu Farha, Gianpiero Francesca, and Juergen Gall

ECCV 2024European Conference on Computer Vision (ECCV)* Equal contribution

Paper Code BibTeX

@inproceedings{Zatsarynna2024gtd,
  title = {Gated Temporal Diffusion for Stochastic Long-Term Dense Anticipation},
  author = {Zatsarynna*, Olga and Bahrami*, Emad and Farha, Yazan Abu and Francesca, Gianpiero and Gall, Juergen},
  year = {2024},
  booktitle = {European Conference on Computer Vision (ECCV)},
  pdf = {https://arxiv.org/pdf/2407.11954},
  code = {https://github.com/olga-zats/GTDA},
  abbr = {ECCV},
  selected = {true},
  preview = {GTD_Icon.png}
}

How Much Temporal Long-Term Context is Needed for Action Segmentation?

Emad Bahrami, Gianpiero Francesca, and Juergen Gall

ICCV 2023IEEE International Conference on Computer Vision (ICCV)

Paper Code BibTeX

@inproceedings{Bahrami2023ltc,
  title = {How Much Temporal Long-Term Context is Needed for Action Segmentation?},
  author = {Bahrami, Emad and Francesca, Gianpiero and Gall, Juergen},
  year = {2023},
  booktitle = {IEEE International Conference on Computer Vision (ICCV)},
  pdf = {https://arxiv.org/abs/2308.11358},
  code = {https://github.com/LTContext/LTContext},
  abbr = {ICCV},
  selected = {true},
  preview = {LTC_Icon.png}
}

TaylorSwiftNet: Taylor Driven Temporal Modeling for Swift Future Frame Prediction

Saber Pourheydari*, Emad Bahrami*, Mohsen Fayyaz*, Gianpiero Francesca, Mehdi Noroozi, and Juergen Gall

BMVC 2022British Machine Vision Conference (BMVC)* Equal contribution

Paper Project Code BibTeX

@inproceedings{Pourheydari2022taylorswift,
  author = {Pourheydari*, Saber and Bahrami*, Emad and Fayyaz*, Mohsen and Francesca, Gianpiero and Noroozi, Mehdi and Gall, Juergen},
  title = {TaylorSwiftNet: Taylor Driven Temporal Modeling for Swift Future Frame Prediction},
  abbr = {BMVC},
  year = {2022},
  booktitle = {British Machine Vision Conference (BMVC)},
  html = {https://taylorswiftnet.github.io/},
  code = {https://github.com/TaylorSwiftNet/TaylorSwiftNet},
  pdf = {https://arxiv.org/pdf/2110.14392.pdf},
  selected = {true},
  preview = {TaylorSwiftNet.png}
}

3D CNNs With Adaptive Temporal Feature Resolutions

Mohsen Fayyaz*, Emad Bahrami*, Ali Diba, Mehdi Noroozi, Ehsan Adeli, Luc Van Gool, and 1 more author

CVPR 2021IEEE Conference on Computer Vision and Pattern Recognition (CVPR)* Equal contribution

Paper Project Code BibTeX

@inproceedings{Fayyaz2021atfr,
  author = {Fayyaz*, Mohsen and Bahrami*, Emad and Diba, Ali and Noroozi, Mehdi and Adeli, Ehsan and Van Gool, Luc and Gall, Juergen},
  title = {3D CNNs With Adaptive Temporal Feature Resolutions},
  booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year = {2021},
  html = {https://similarityguidedsampling.github.io/},
  pdf = {https://arxiv.org/pdf/2011.08652.pdf},
  code = {https://github.com/SimilarityGuidedSampling/Similarity-Guided-Sampling},
  selected = {true},
  preview = {ATFR.png},
  abbr = {CVPR}
}