Muhammad Osama Zeeshan

3rd Floor, Pavillon Principal (A), ETS

Montreal, Canada

👋 Welcome to my academic and professional journey!

Open to work and consultancy opportunities in machine learning, computer vision research, and applied scientist roles.

I am a developer, consultant, ML researcher, and doctoral candidate based in Montreal, Canada, with several years of experience across industry and research. My work focuses on real-time, robust computer vision systems that combine deep learning, domain adaptation, multimodal learning, VLMs, and LLMs to extract, localize, and recognize patterns in visual data.

🌱 I am currently pursuing a Ph.D. on Adaptation of Deep Learning Models for Video-Based Expression Recognition under the supervision of Prof. Eric Granger and Prof. Marco Pedersoli.

🏆 FRQNT Scholar

What I Build

Research

Personalized Computer Vision

Unsupervised domain adaptation, multimodal learning, VLMs, and LLMs for personalized visual understanding.

Applied CV

Real-World Vision Systems

OCR, medical imaging, solar analysis, and video understanding pipelines built for noisy real-world data.

Engineering

ML & Software Delivery

Python/PyTorch experiments, RESTful APIs, backend systems, reproducible pipelines, and production-ready tools.

News

Jun 21, 2026	Our paper “CLIP-AUTT: Test-Time Personalization with Action Unit Prompting for Fine-Grained Video Emotion Recognition” has been accepted to ECCV 2026: The 19th European Conference on Computer Vision, Malmö, Sweden. Read more
May 16, 2026	Two of our papers have been accepted to ICLR 2026: BAH Dataset for Ambivalence/Hesitancy Recognition in Videos for Digital Behavioural Change and Personalized Feature Translation for Expression Recognition: An Efficient Source-Free Domain Adaptation Method.
Feb 01, 2026	Started as a Visiting Research Scientist with STARTS, Inria - French National Institute for Research in Digital Science and Technology in Valbonne, France, working on computationally efficient ML methods for personalized human behavior and expression analysis, test-time multimodal domain adaptation, and emerging AI/ML methods for efficient personalized modeling.
Jan 01, 2026	Our paper “MuSACo: Multimodal Subject-Specific Selection and Adaptation for Expression Recognition with Co-Training” has been published in IEEE/CVF Winter Conference on Applications of Computer Vision (WACV) 2026. Read more
Jun 01, 2025	Our paper “Progressive Multi-Source Domain Adaptation for Personalized Facial Expression Recognition” has been published in IEEE Transactions on Affective Computing. Read more

Selected Publications

CLIP-AUTT: Test-Time Personalization with Action Unit Prompting for Fine-Grained Video Emotion Recognition

Muhammad Osama Zeeshan , Masoumeh Sharafi , Benoît Savary , and 3 more authors

In ECCV 2026: The 19th European Conference on Computer Vision, Malmö, Sweden , 2026

PDF
MuSACo: Multimodal Subject-Specific Selection and Adaptation for Expression Recognition with Co-Training

Muhammad Osama Zeeshan , Natacha Gillet , Alessandro Lameiras Koerich , and 3 more authors

In WACV 2026: IEEE Winter Conf. on Applications of Computer Vision, Tucson, Arizona, USA , 2026
Progressive Multi-Source Domain Adaptation for Personalized Facial Expression Recognition

Muhammad Osama Zeeshan , Marco Pedersoli , Alessandro Lameiras Koerich , and 1 more author

IEEE Transactions on Affective Computing, 2025

Subject-based domain adaptation for facial expression recognition

Muhammad Osama Zeeshan , Muhammad Haseeb Aslam , Soufiane Belharbi , and 4 more authors

In 2024 IEEE 18th International Conference on Automatic Face and Gesture Recognition (FG) , 2024

Bib

@inproceedings{zeeshan2024subject,
  title = {Subject-based domain adaptation for facial expression recognition},
  author = {Zeeshan, Muhammad Osama and Aslam, Muhammad Haseeb and Belharbi, Soufiane and Koerich, Alessandro Lameiras and Pedersoli, Marco and Bacon, Simon and Granger, Eric},
  booktitle = {2024 IEEE 18th International Conference on Automatic Face and Gesture Recognition (FG)},
  pages = {1--10},
  year = {2024},
  organization = {IEEE}
}