Human activity recognition has gained popularity in the last two decades. Numerous applications with various concepts and trends have been developed as a result of the most recent advancements in digital world technology. The most rigorous area of computer vision research in today's trending world is Human Activity Recognition, a machine learning technique that is widely used in emerging contexts like healthcare maintenance, safety, reliability, and human–computer interaction. The study focuses on the advancements in activity detection methodologies, particularly activity representation and classification methods. The representation approaches and generative models for classification methods evaluate numerous popular methods. The aim is to discuss the functioning of the model that exhibits maximum accuracy and performance in detecting the actions and give an overview of such strategies.