My research spans computer vision, representation learning, and system-level evaluation. I’m interested in how representations and system design shape real-world model behaviour. I also explore multimodal AI and 3D spatial AI.
Semantic View Induction for Object-Guided Scene Structuring
Introduces a data-centric framework for discovering recurring visual configurations in scene datasets and using open-set object evidence to organise samples without changing category labels or discarding ambiguous cases.
@inproceedings{Bamigbade2027SceneStructuring,
author = {Bamigbade, Opeyemi and Scanlon, Mark and Sheppard, John},
title = {Semantic View Induction for Object-Guided Scene Structuring},
booktitle = {Proceedings of the Computer Vision Conference (CVC) 2027},
year = {2027},
publisher = {Springer Nature Switzerland},
address = {Cham}
}
VAAS: Vision-Attention Anomaly Scoring for Image Manipulation Detection in Digital Forensics
Introduces a dual-module approach combining global attention-based anomaly estimation with patch-level self-consistency scoring, formalising attention-driven representation analysis as a structured mechanism for interpretable anomaly detection.
Journal: Forensic Science International: Digital Investigation · DFRWS EU · 2026
@article{Bamigbade2026VAAS,
author = {Bamigbade, Opeyemi and Scanlon, Mark and Sheppard, John},
title = {VAAS: Vision-Attention Anomaly Scoring for Image Manipulation Detection in Digital Forensics},
journal = {Forensic Science International: Digital Investigation},
volume = {56},
pages = {302063},
year = {2026},
doi = {10.1016/j.fsidi.2026.302063}
}
Improving Image Embeddings with Colour Features in Indoor Scene Geolocation
Studies colour-aware representation learning, combining dominant colour descriptors and colour statistics with visual embeddings for indoor scene understanding.
@article{Bamigbade2025ColourFeatures,
author = {Bamigbade, Opeyemi and Scanlon, Mark and Sheppard, John},
title = {Improving Image Embeddings with Colour Features in Indoor Scene Geolocation},
journal = {IEEE Access},
volume = {13},
pages = {79860--79870},
year = {2025},
doi = {10.1109/ACCESS.2025.3564496}
}
Computer Vision for Multimedia Geolocation: A Systematic Literature Review
Synthesises 123 studies on computer vision and AI for multimedia geolocation, identifying methodological gaps and design challenges for robust real-world systems.
@article{Bamigbade2024MultimediaGeolocation,
author = {Bamigbade, Opeyemi and Sheppard, John and Scanlon, Mark},
title = {Computer Vision for Multimedia Geolocation in Human Trafficking Investigation: A Systematic Literature Review},
journal = {arXiv preprint arXiv:2402.15448},
year = {2024},
doi = {10.48550/arXiv.2402.15448}
}
Peer reviewing
I review work in AI, machine learning, and computer vision, with attention to technical rigor, reproducibility, evaluation quality, and whether methods can hold up in practical settings.
