Human Action Recognition – Analysis of Deep Learning Methods

Human Action Recognition (HAR) is significant in many areas, including surveillance security, healthcare services, and industry monitoring. HAR systems automatically learn the data captured by sensors or cameras and provide information regarding the activities performed in an environment. The evolution of deep learning techniques has substantially increased the recognition capabilities of Human Action Recognition (HAR) systems. This study seeks to examine different deep learning methodologies applied to HAR systems. A thorough evaluation of the implemented techniques, their effectiveness, and the distinct benefits associated with each approach is also performed. The study performs a detailed analysis of deep learning techniques employed in recent HAR systems to identify emerging research trends. It examines the challenges and constraints of each approach to uncover existing research gaps. The paper also makes a comparative analysis of methods using different performance metrics to get a clear idea of the efficiency of the approaches discussed. The research emphasizes predominance of hybrid deep learning methodologies in the domain of Human Action Recognition.