Our Research

Humans and other animals use vision as their primary sense for extracting information about the world. The goal of the Visual Geometry Group (VGG) is to build this sense of vision in machine and artificial intelligence. The impact of this technology is vast: besides obvious applications such as robotics, computer vision can be used in any application where images and videos must be analysed with more precision, repeatability, and scale than humans can. Examples include searching large image and video collections such as the Internet or the BBC news archives for content; production and quality control in industrial processes; statistical analysis of large corpora of artefacts such as collections in archeology, sculpture or fine art; analysis of camera traps and other imagery of animals in the wild to build large scale statistics on behaviour or animal welfare; or detecting problematic content such as illegal pornography, hate or abuse in social networks.

The Visual Geometry Group is a recognised international leader in this area, having pioneered many of the techniques now in widespread use in the community: in multiple-view geometry, image retrieval (the VGG invented the first `Google for images'), recognition in videos and, more recently, deep learning, where it introduced some of the most widely used models in the area.

The group has numerous industrial collaborations, including with the BBC, Continental corp, Toshiba, Nielsen and many others.

AI  -  Computer Vision  -  Machine Learning  -  Deep Learning  -  Visual Recognition