Home (original) (raw)
I am Professor of Computer Vision and Machine Learning and a co-lead of theVGG group at the Engineering Science department of the University of Oxford. I researchcomputer vision and machine learning methods with a focus on Spatial Intelligence.
If you are looking for aPhD (DPhil) or postdoctoral (PDRA) position, please see here.
VGGT-Ω is a feed-forward 3D reconstruction model that scales reconstruction quality with data and model size, while also improving efficiency and dynamic-scene performance. Compared to the original VGGT model, it introduces architectural changes such as register-based information exchange and a simplified dense prediction head, enabling stronger results with lower training memory.
VGGT-Ω demo: reconstruction of a complex video (courtesy of M. Parkhomenko).