Visual Geometry Group

Visual Geometry Group

The Visual Geometry Group is a research collective based at the Department of Engineering Science, University of Oxford, known for advances in computer vision, machine learning, and image understanding. It has produced influential datasets, algorithms, and software that have shaped work at institutions such as Stanford University, Massachusetts Institute of Technology, Carnegie Mellon University, and Google Research. The group’s output is frequently cited alongside work from Microsoft Research, Facebook AI Research, DeepMind, and OpenAI.

History

The group was founded in the mid-1990s within the University of Oxford and has interacted with institutions including Imperial College London, University College London, Princeton University, Harvard University, Yale University, Columbia University, and California Institute of Technology. Early milestones occurred contemporaneously with developments at Bell Labs, Xerox PARC, AT&T Laboratories, and IBM Research. Over time the group’s work paralleled breakthroughs from researchers at the National Institute of Standards and Technology, European Laboratory for Particle Physics, and Max Planck Institute for Informatics. Funders and partners have included the Engineering and Physical Sciences Research Council, European Research Council, Wellcome Trust, and Royal Society.

Research Areas

Research spans core topics in computer vision such as object recognition, image classification, and feature descriptors, connecting to subfields touched by researchers at Stanford AI Lab, MIT Computer Science and Artificial Intelligence Laboratory, and Berkeley Artificial Intelligence Research. The group investigates deep learning architectures similar to work at Google Brain, Microsoft Research Cambridge, and Nvidia Research, and explores geometric computer vision themes studied at ETH Zurich, INRIA, and Technical University of Munich. Other linked areas include 3D reconstruction studied at Facebook Reality Labs and Oculus Research, structure-from-motion topics associated with UCLA and Brown University, and visual SLAM research explored at Oxford Robotics Institute and University of Pennsylvania GRASP Lab.

Notable Projects and Contributions

The group has released influential datasets and models that are often referenced alongside ImageNet, COCO, PASCAL VOC, and KITTI benchmarks from institutions such as Princeton and Toyota Technological Institute at Chicago. Their descriptors and convolutional network architectures are cited in literature from Google Scholar entries tied to researchers at Cornell University, University of Washington, and University of Cambridge. Software released by the group is used in pipelines developed at Adobe Research, Apple Machine Learning Research, Amazon Research, and Samsung Research. The group’s benchmarks have been evaluated in challenges organized by NeurIPS, CVPR, ICCV, ECCV, and ICLR, and recognized in awards from ACM, IEEE, Royal Academy of Engineering, and British Computer Society.

People and Leadership

Leadership and alumni have included figures who have collaborated with scholars at Oxford, Cambridge, and Harvard, and who later joined institutions such as DeepMind, Google, Microsoft, Amazon, and Meta. The group’s members have coauthored papers with researchers from Princeton, Stanford, Caltech, and ETH Zurich, and have participated in panels at the Royal Society, IEEE, ACM, and European Commission workshops. Postdoctoral researchers and doctoral students have moved to roles at Carnegie Mellon University, University of Toronto, University of British Columbia, and Peking University, and have contributed to consortia including Alan Turing Institute and European Space Agency projects.

Facilities and Collaborations

Physical and computational facilities supporting the group are hosted within the University of Oxford and have interfaced with national computational resources such as ARCHER, UK Research and Innovation computing infrastructure, and European supercomputing centers used by researchers from CERN and EMBL. Collaborations extend to industry labs including Google DeepMind, Microsoft Research, Facebook AI Research, Amazon Science, Baidu Research, and Tencent AI Lab, and to governmental and non-governmental organizations like the Royal Society, European Commission, and Wellcome Trust. The group participates in networks with institutions such as King's College London, University of Southampton, University of Edinburgh, and University of Manchester, and in international partnerships involving Stanford, MIT, and Carnegie Mellon.

Category:Computer vision research groups