Human Activity Identification Using Image Processing
P. Anupriya, D Mythili · International Journal of Research Publication and Reviews · 2025
Human Activity Recognition (HAR) has gained significant attention in recent years due to its wide-ranging applications in healthcare, sports analytics, and context-aware computing.Convolutional Neural Networks (CNNs) have demonstrated remarkable success in various computer vision tasks, prompting exploration into their efficacy for HAR.This paper presents a comprehensive exploration of CNN-based approaches for HAR, encompassing data collection, preprocessing, model architecture design, training strategies, evaluation methodologies, and deployment considerations.We delve into the intricacies of CNN architectures tailored for HAR, discussing the integration of convolutional layers for feature extraction from raw sensor data and subsequent layers for activity classification.Furthermore, we investigate the challenges inherent in CNN-based HAR, including the scarcity of labeled data, model interpretability, and realtime inference constraints.We analyze recent advancements in CNN-based HAR techniques, such as transfer learning and attention mechanisms, and their impact on performance.Additionally, we discuss the implications of deploying CNN models in real-world scenarios, emphasizing considerations related to computational efficiency and scalability.Through this deep exploration, we aim to provide researchers and practitioners with a comprehensive understanding of the state-of-the-art in CNN-based HAR and inspire future innovations in this rapidly evolving field.