Brighton academics today (1 February) launched a nationwide network of researchers developing technology which will enable computers not only to see the world and to communicate using language, but also understand what they see, hear and say.
Example applications include software tools that speak and describe to visually impaired internet users what is on their screens and make it easier for them to use websites. Other innovations include tools that can save police valuable time by automatically sifting through CCTV footage to find specific types of crimes.
Dr Anja Belz at the University of Brighton's School of Computing Mathematical and Information Sciences has been chosen to head the EPSRC Network on Vision and Language which is funded by the UK's Engineering and Physical Sciences Research Council (EPSRC).
The network will run for three years and will receive 120,000 in funding for events including annual network meetings, conferences and exhibitions, including a series of evening lectures that will be open to the public.
Dr Belz said: "Our role will be to pump-prime research that is focussed on combining computer vision and language, by helping researchers develop their ideas into viable project proposals. In addition to offering financial support, we will provide online resources such as a searchable database of experts and repositories of downloadable data, software and publications.
"This is a prestigious award and we are very excited by the project and look forward to stimulating research in this field."
Anja Belz and her team won an EPSRC competition to lead the network which already has some 100 UK-based member researchers, and will develop extensive links to researchers all over the world.
Helping the visually impaired is a key research aim. A study by accessibility experts Nomensa found that three-quarters of Britain's top FTSE 100 companies failed to meet minimum accessibility requirements as laid down by UK disability discrimination legislation and World Wide Web Consortium guidelines. Some companies and organisations around the world have been successfully sued.
Dr Belz said: "The Royal National Institute for the Blind compares the web experience of visually impaired users to visiting your favourite shop, finding it open but its doors locked to you while open to others.
"Network members are working towards software tools such as automatic image description tools that can help impaired users gain some degree of access to all visual online content."
Another research field concerns the 4.2 million CCTV cameras now operating in the UK.
Dr Belz said: "Police have an ever-growing mountain of potential visual evidence which could help solve crime but in practice, according to the London Metropolitan Police, they do not perform the tedious job of sifting through footage for any but the most serious crimes.
"For example, only three per cent of London's street robberies are currently being solved using security cameras."
Dr Belz added: "Some of our members are working towards software tools capable of sifting through CCTV footage automatically and locating sequences matching a given type of crime."