Not everyone knows that each person has an individual style of walking. Gait each unique, like a snowflake, retina or fingerprint. Person to see the differences is difficult, but artificial intelligence is able to learn it.
According to a new study, published on IEEE Transactions on Pattern Analysis and Machine Intelligence, the new system, dubbed StoofBD, capable of almost accurately recognize people by gait. It is 380 times more accurate than previously developed system and operates autonomously.
StoofBD was developed at the University of Manchester. The Creator claims that while walking we use 24 different movements, which make up a unique gait. To create the recognition system the database was created from 20 thousand steps 120 people. Each step was measured and photographed in high resolution.
In addition, each step was measured using a special carpet. Information from these rugs handles artificial intelligence, finding of conformity. This is what makes the system so accurate.
It is assumed that StoofBD can be used for security purposes in airports and at home. In addition, the system can be used to control activities in the workplace. For equipment you will need to cover the floor of a special carpet and to equip the premises with cameras with high resolution. It is quite expensive, and not very clear advantages over facial recognition system, which is already widely used in different parts of the world.