Recent Papers

Human Activity Recognition (HAR) has emerged as a critical research area with wide-ranging applications in healthcare monitoring, smart homes, humancomputer interaction, and security. Traditional machine learning methods relied on handcrafted features, which often struggled with variability in human motion, environmental conditions, and sensor modalities. In recent years, deep learning has revolutionized HAR by enabling automatic feature extraction and improved performance across diverse datasets. This review paper provides a concise overview of state-of-the-art deep learning approaches for HAR, focusing on Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), hybrid architectures, and the recent adoption of attention mechanisms and transformer models. We also highlight multimodal fusion strategies that integrate wearable sensors, vision data, and ambient sensing for more robust recognition. Beyond summarizing existing techniques, this review critically analyzes current challenges such as data scarcity, computational costs, model generalization, and privacy concerns. We further outline emerging research directions, including lightweight models for edge devices, transfer and self-supervised learning, explainable HAR, and privacy-preserving frameworks. By consolidating recent advances and open issues, this paper aims to guide future research efforts toward more accurate, efficient, and ethical HAR systems powered by deep learning

Copyright © 2025 Jitendra Lakhawat. This is an open access article distributed under the Creative Commons Attribution License.