Toe Walking Detector

Co-creator and Developer

PERIOD

2019

PARTNER

Luiz Freitas

TWD is a software/hardware platform that aims to assist with gait analysis. More precisely, TWD can detect when an individual is toe walking instead of regular heel-to-toe walking. It uses Machine Learning to classify between the walking types.

THE TWD

The Toe Walking Detector (TWD) is a software and hardware platform that aims to assist researchers with gait analysis. More precisely, TWD can detect when an individual is toe walking instead of regular heel-to-toe walking.

TWD is composed of two parts:

  1. A small IMU sensor that can be installed on the individual's shoe, allowing it to collect its feet motion data, like linear acceleration and angular velocity;
  2. The software that uses a Convolutional Neural Network to process the sensor motion data and correctly classify between regular and toe walking. By now, we're getting a classification accuracy higher than 85%.

WHAT I DID

This project was co-created by psychologist Luiz Freitas and me. I'm working as the Machine Learning engineer and software developer.

We're using Python to develop all the software (Desktop with PyQt and Jupyter Notebooks). We're also using Keras to train and run the CNNs.