Human Activity Detection Using Deep Learning

Human Activity Recognization (HAR), which focuses mostly on behavioral patterns, security, and health monitoring, is one of the crucial fields of machine learning. The focus is on detecting human activities like eating, sleeping, and watching by using deep learning techniques on a dataset of fourteen classes. Due to their demonstrated accuracy and efficiency, CNNs (Convolutional Neural Network) are used in the basic architecture for processing image-based data, coupled with EfficientNetB0, a more current state-of-the-art design that strikes a compromise between computational efficiency and performance. Overfitting and model optimization issues are addressed while maintaining accuracy. With an accuracy of 87.50 %, this architecture demonstrates that deep learning is an effective tool for detecting human activity. This study provides a thorough description of the techniques employed.