New technology helps identify people through gait

Our individual walks are like snowflakes, unique and non-overlapping.

Understanding this, technology scientists have developed a new and efficient system that helps identify people through gait , using artificial intelligence. And in theory, it could replace the retina scanner, replacing fingerprint recognition technology and maybe the security control stations we're using.

This system can see specific points in a person's gait, and use those points to identify people with near-perfect accuracy - according to a new study published in IEEE magazine. Transactions on Pattern Analysis and Machine Intelligence (a journal published by the IEEE computer community). The new system, called SfootBD, can deliver 380 times more accuracy than previous methods, and it doesn't require people to "walk barefoot to function properly . " It has less impact on people than other biometric-based authentication systems, such as retinal scans or fingerprints, but this is a major concern about rights. privacy - because it can be used secretly.

This system can see the special features of a person's gait.

"Every human being has about 24 different elements and movements when moving, so each different person will have different gait and that gait is unique" - Omar Costilla Reyes - the lead author of the study. This new rescue - and a Manchester University technology scientist said in an interview.

According to Gizmodo, to create the SfootBD system , Reyes has built a database of 20,000 footsteps signals from more than 120 different people. So far, this is still the largest database of human footsteps. Each gait is defined by a pressure gauge mounted on the floor and a high resolution camera. A system uses the location to collect those data, analyze the weight distribution through each step, speed and analysis of the three-dimensional space of each step. It is important that the aforementioned AI system carefully reviews every aspect of the gait, rather than focusing on the shape of the footprint.

The AI ​​system will carefully consider every aspect of the gait, rather than focusing on the shape of the footprint left behind.

Reyes said: "It is very difficult to focus on the way of walking by tracking the force on the floor in one step. By distinguishing different forces from the footsteps of different people. And that's why we have studied this new AI system to help solve this challenge from a new and potential perspective. "

Previously, scientists could only identify the footsteps of people when they did not wear shoes, and now, we have a 3D imaging technology that compares a person's gait to a mechanical system. database of gait. This new technology offers greater accuracy, although it requires the use of pressure pads mounted on the floor.

To test the SfootBD system, Reyes' research team conducted follow-up participants in different contexts: at airport security checkpoints, at workplaces, and at their homes. Researchers also tested a group of people who were asked to impersonate others to see if the system's artificial intelligence could be detected if someone intentionally imitated the footsteps or gaits of others. no (if possible imitate).

The results showed that the system achieved an average accuracy of 100% when identifying individuals, the error rate was 0.7%. This is clearly an amazing result, and is a sign that high technology is very useful and effective in our everyday life.

The system achieves an average accuracy of 100% when identifying individuals.

This new system also has some limitations. As mentioned above, SfootBD requires the use of floor pads with pressure measurement and a high resolution camera, so it is not usable everywhere. Moreover, SfootBD only works strongly when it has a large database, because the system only identifies those who have been tracked, recorded and stored in the system database. Such an approach is not very flexible, and scalability is not high, because collecting and storing people's gait "characteristics" in order to be more difficult than storing those face recognition photo. Finally, as mentioned above, it is still a matter of privacy when this technology can be used secretly and stealthily to follow others without them knowing.

Clearly, this is a remarkable progress in the current technology of artificial intelligence. But only time will prove the practical application of this method in our real world.