A Comparative Study of CNN and RNN Architectures For Human Activity Recognition Using Wearable Sensors

This study presents a comparative analysis of Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs) for Human Activity Recognition (HAR) using data from wearable sensors. HAR is crucial in various applications such as healthcare, fitness tracking, and smart environments, where accurate recognition of human activities is essential. While traditional methods have been used in HAR, the emergence of deep learning has significantly improved performance. In this work, we implemented and evaluated both CNN and RNN models on a benchmark HAR dataset, focusing on key metrics such as accuracy, precision, recall, and F1-score. The results demonstrate that CNNs outperform RNNs in terms of overall accuracy and reliability, particularly in activities characterized by distinct spatial patterns. However, RNNs remain competitive in recognizing activities with complex temporal dynamics. These findings highlight the strengths and limitations of each architecture and provide insights into their suitability for different types of HAR tasks.