Human activity recognition with the support of deep learning and machine learning-survey
Sunitha Sabbu, Vithya Ganesan · IET conference proceedings. · 2023
Human Activity Recognition (HAR) is a method of automatically recognizing human actions based on streaming data from a variety of sensors, including inertial sensors, physiological sensors, position sensors, camera, time, and a variety of other ambient sensors. HAR has been shown to be beneficial in a variety of fields of study, including healthcare, elderly care, ambient living, personal care, social science, rehabilitation engineering, and a variety of other fields. Deep learning-based algorithms have become the most effective and efficient choice of algorithms for recognizing and solving HAR problems, thanks to recent advances in processing power. In this survey, we use deep learning algorithms to categories recent research work according to numerous variables and measurements in order to analyses recent trends in HAR. The articles are examined in terms of HAR, time series analysis, machine learning models, dataset construction methods, and the application of various other emerging trends such as transfer learning, active learning, and so on.