Evaluating Machine Learning Models for Age Group Classification Through Human Gait

The human gait is a complex process that is continued to be studied today. The pattern in which every individual moves is one that is unique to that person and is affected by many factors. It was decided to explore if there were certain human factors that influence differences in human gait. As age was a commonly studied natural factor that alters the human gait, this research attempted to see if an artificial intelligence would be able to classify people into different age groups depending on their gait. In our research it was found that after using pose estimation technology to obtain joint angles of different parts of the body during gait, an artificial intelligence is able to classify people into different age groups by analyzing their gait. The final model used was an XGBoost model that during testing yielded a final accuracy of 96%. Through these findings it is hoped that this discovery assists further study of the human gait and the factors that affect its patterns.