Human activity recognition (HAR) has garnered significant attention in recent years, particularly with the advent of deep learning techniques, which have proven highly effective in recognizing complex tasks. Deep learning approaches offer superior performance and lower costs compared to traditional machine learning methods. This paper provides a comprehensive survey of state-of-the-art HAR models that utilize deep learning architectures, including Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), and hybrid systems combining multiple architectures. The analysis explores the implementation strategies of these models to enhance their effectiveness and discusses potential limitations and challenges that they may encounter.

Human Activity Recognition, Deep Learning, Convolutional Neural Networks (CNN), Long Short-Term Memory (LSTM), Hybrid Systems