The three-year project aims to adapt artificial intelligence software already being developed by researchers at the University’s Institute of Industrial Research to identify visual patterns. The software is so sophisticated it can identify minor visual cues such as whether a car aerial is up or if the car has a dent as well as more complex cues such as violent behaviour.
If, for example, a CCTV camera is trained on a shopping precinct, the software would use artificial intelligence to notice if someone raises their arm suddenly or runs. The operator of the camera has banks of monitors to watch but the new software will alert them immediately to any unusual activity and, if necessary, call the police.
The research team is now working on using the same software to ‘learn’ sounds and react to them by swinging the CCTV camera towards in them at the same speed a person would turn their head if they heard someone scream, which is about 300 milliseconds.
Dr David Brown, director of the Institute, said: “The visual-recognition software will be able to identify visual patterns but for the next stage we want to get the camera to pivot if it hears a certain type of sound. So, if in a car park someone smashes a window, the camera would turn to look at them and the camera operator would be alerted.”
“The longer artificial intelligence is in the software the more it learns. Later versions will get cleverer as time goes on, perhaps eventually being able to identify specific words being said or violent sounds. We are only listening for specific words associated with violence, not full conversations,” added Brown.
The software behind this research uses fuzzy logic to identify certain visual cues and sounds. Dr Brown said, “In identifying sound, we are looking for the shapes of sound. In the same way, if you close your eyes, you can trace the shape of a physical object and ‘read’ its profile with your hand we are developing shapes of sound so the software recognises them.
“The software will use an artificial intelligence template for the waveform of sound shapes and if the shape isn’t an exact fit, use fuzzy logic to determine what the sound it. For example, different types of glass will all have slightly different waveforms of sound when they smash but they will have the same generic shape which can be read using fuzzy logic. It’s a very fast, real-time method of identifying sounds.”
Besides alerting camera operators to potentially threatening behaviour and crimes in action, the software will mean police officers no longer have to sit through hours of recordings to identify a specific object, action or noise after a crime is known to have taken place because the software could be used to identify specific objects.
The technology would also overcome the problem of camera operators losing concentration when watching banks of television screens showing normal activity because it would alert them in real time to threatening behaviour and sounds.
By the end of the three-year programme, the researchers hope to have generated algorithms that can be used inside existing CCTV software. And because the system is underpinned by artificial intelligence, each successive generation of algorithms would become more sophisticated as they ‘learn’ what they are looking and listening out for.
The new research is being funded by the Engineering and Physical Sciences Research Council (EPSRC). |
