Recognition of Human Activity using Signal Processing & Deep Neural Networks

The task of tracking the activities being performed by humans is of immense importance as it finds application in healthcare, marketing, surveillance, etc. To keep track of this, they need to be recognized and classified into various activities. The data can be collected through smart devices such as smartwatches or smartphones which internally use various sensors to measure the physical movements of the person using it. We have used Human Activity Recognition Dataset from the UCI ML Repository, which consists of the data being collected from the smartphones of 30 volunteers. After choosing the appropriate features, we trained an Artificial Neural Network and a Long Short Term Memory Neural Network to classify the activities into various categories and performed a comparative analysis of these neural networks among themselves & also with respect to the previously leveraged techniques & algorithms on the dataset.