Human activity recognition system
Divaksh Parmar, Mitanshu Bhardwaj, Aayush Garg, Anjali Kapoor, Anju Mishra · 2023
Smartphones offer a great platform for ongoing behavioral research. However, it might be difficult to observe people's behavior in the wild because users' actions and behaviors can vary depending on the situations and environment's they are in. The user's behavioral environment must be taken into account when modelling and analyzing human activity in the wild because it is just as important as knowing the variety of physical activities. This study integrates human behavioral circumstances with physical activities to create a unique framework for context-aware human activity recognition. The goal of the current active research area called "Human Activity Recognition" is to comprehend human behavior through the interpretation of sensory data. The recordings of 30 study participants engaging in activities of daily living (ADL) while wearing a smartphone strapped on their waist with inertial sensors were used to create the Human Activity Recognition database. Each individual carried out the fundamental tasks of (WALKING, STAIRS CLIMBING, SITTING, STANDING, SLEEPING). Different classification techniques, including CNN and Random Forest, were used to construct the system. The outcome of the experiment demonstrates the precision, ability to distinguish between activities, and ability to graphically compare data.