This Machine Learning Algorithm Identifies you by your Walking Style

Aishwarya Singh 10 May, 2019 • 2 min read

Overview

  • This AI system can identify people based on their gait and walking styles
  • The researchers collected data from 127 unique individuals; this included approximately 20,000 footstep signals
  • The final model gave an accuracy of almost 100%! (There was a 0.7 error rate)

 

Introduction

The only ways of practically identifying people was either by fingerprint scanning or using CCTV camera to directly identify them. Recently with the advancements in machine learning, we have seen facial recognition software take a giant leap forward. But these have known issues which crop up from time to time.

So here comes a system, powered by Artificial Intelligence, that can pick out an individual based on the way they walk!

This artificial intelligence system is developed by a group of researchers from the University of Manchester and the University of Madrid. The main idea behind the study was to train their model to distinguish between various walking speeds and styles. According to one of the researchers, each individual has 24 unique factors and movements when they are walking, and these can be used as a benchmark for identifying them. This is essentially the equivalent of scanning fingerprints or the retina.

In order to build an the AI, the team first needed to collect a lot of data that would eventually help them distinguish walking styles. So the team collected a footstep database consisting of approximately 20,000 footstep signals from 127 people. Using floor-only sensors and high-resolution cameras, they compiled the samples and the dataset. This dataset, called SfootBD, was used to develop the advanced computational models needed for automatic footprint biometric verification.

To validate the final model, the researchers put it to the test in three distinct security scenarios. According to the scientists, their AI system correctly identified a person almost 100 percent of the time, with an error rate of just 0.7 percent.

According to the Financial Express, footprint recognition is an ideal process because it is ultimately non-intrusive to the person being verified by the AI. The individual doesn’t even need to bother removing them footwear – the algorithm uses their gait to identify them.

 

Our take on this

The most obvious application of this technology is for security purposes. Since the accuracy of the AI system is very close to 100%, it definitely bears testing out in practical scenarios to see how it holds up. Although this is a creative idea and will save a lot of time for an individual (since it’s non intrusive), I personally have one concern – will this system accurately identify an individual when he/she changes his pace, for instance while running to catch a flight?

Regardless, this is an excellent use of data and machine learning. As a data scientist, you need to think outside the box and come up with use cases never seen before, even on scenarios that most people feel have no scope of innovation.

 

Subscribe to AVBytes here to get regular data science, machine learning and AI updates in your inbox!

 

Aishwarya Singh 10 May 2019

An avid reader and blogger who loves exploring the endless world of data science and artificial intelligence. Fascinated by the limitless applications of ML and AI; eager to learn and discover the depths of data science.

Frequently Asked Questions

Lorem ipsum dolor sit amet, consectetur adipiscing elit,

Responses From Readers

Clear

Pranov Shobhan Mishra
Pranov Shobhan Mishra 29 May, 2018

Great article, great idea. The machine learnt predicting by being fed with walking movements of known individuals and labelling. Will it be able to predict or recognize if an unknown person is in scope? i guess not. regardless it can still be useful if someone unknown is in the area of scope.

Erik Lee
Erik Lee 30 May, 2018

Great Article! I love the last sentence you said, even most people feel have no idea of innovation, data scientists are still trying their best to make the impossible possible, and at many times they succeed.