Human Activity Recognition And Fall Detection Using Sensor-Based Data And Machine Learning Models
Authors
Vinod, Athira
Issue Date
2025.17.12
Degree
Master of Business Administration
Publisher
Dublin Business School
Rights holder
Rights
Open Access
Abstract
A large number of old people, and people with physical ailments need to be monitored as they may not be able to control different parts of their body, which leads to problems as these people may fall or end up in some dangerous other situation. These people need to be constantly monitored to ensure their safety, however manually monitoring these people can be a tedious process. In the study proposed here, a prototype system is proposed for Human Activity Recognition(HAR), and early fall detection. The study uses two machine learning models, the Time-Series Transformer (TST), and the Random Forest(RF) classifier. These models were trained using data collected by sensors obtained from a dataset. The TST and RF showed good performances in HAR, and early fall detection. These two models were integrated to a web app prototype which acted as the system that performed both HAR, and early fall detection