FRONTIER SCIENCE AND TECHNOLOGY
years of 30 THINKING MACHINES
SURREY’S CENTRE FOR VISION, SPEECH AND SIGNAL PROCESSING (CVSSP) CELEBRATED ITS 30TH ANNIVERSARY THIS YEAR, MARKING THREE DECADES OF LEADING SCIENTIFIC ADVANCES IN AI AND MACHINE PERCEPTION FOR THE BENEFIT OF SOCIETY.
Having pioneered the concept of ‘pattern recognition’ back in the 1980s, the Centre has since created machines that can not only see and hear, but also understand the world around them.
The Centre’s 30th anniversary celebration was an opportunity for academics, researchers and industry partners to reflect on a global research legacy which has, among many other achievements, transformed the way stories are brought to life in films through 3D image capture techniques, and enabled cancer therapies to be more precisely targeted, based on innovative medical imaging.
The only centre of its kind in the UK, CVSSP is playing a vital role in today’s artificial intelligence (AI) revolution not only through its wide-ranging research, but also by training the highly skilled workforce which will lead development of technologies to enrich, improve and even save lives in the future.
CHANGING THE FACE OF SECURITY
Machines which can recognise faces have become a crucial tool in keep society safe, but what happens when an image is taken in poor light or in a crowded environment, such as on CCTV footage?
CVSSP’s biometrics experts are working on a solution to this challenge, which will mark a step-change in face-matching technology. In collaboration with the Home Office’s Centre for Applied Science through the £6m FACER2VM project, they are developing AI-driven ‘unconstrained’ facial recognition systems which can reliably identify people in difficult conditions such as in variable lighting, pose, facial expression, and even in cases of identical twins.
The Centre has been researching face recognition since 1995. It exploited machine learning to create 3D face reconstructions from 2D images and video – a technique which was commercialised through spin-out company OmniPerception and is widely used today by UK police forces and border security.
Demonstrating its success, in May 2018 CVSSP’s face recognition system was pitted against a human ‘super recogniser’ (someone with an exceptional gift for remembering and matching faces) on a TV programme on the human brain, produced by Japanese broadcaster NHK – and it outperformed the human by a margin of 20 per cent.
MAKING SENSE OF A SEA OF DATA
Visual recognition is a fundamental capability which we humans take for granted, but a task which computers still struggle to perform quickly and reliably – as anyone who has tried to find images or videos of a specific object online will testify. However, achieving accurate visual recognition is crucial for applications in many emerging areas such as robotics, autonomous vehicles, smart devices and big data.
Over the past five years, CVSSP has developed some of the world’s best-performing machine perception technologies, which are capable of identifying unique objects from vast volumes of data.
One example is the system the Centre has created for the Innovate UK-funded iTravel project: a smartphone-based intelligent ‘virtual journey assistant’ which enables users to get a visual fix on their position and see an augmented view of their surroundings. This technology was put to the test in the Google Landmark Retrieval Challenge 2018 – which challenged systems to automatically identify landmarks and retrieve relevant images from a dataset of over a million – and saw CVSSP ranked first, beating 218 teams from around the world.
30
Forever Surrey 2019
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52
