Human Activity recognition Using Machine Learning

Abstract

Human Activity Recognition (HAR) is a rapidly evolving area of research that focuses on identifying and classifying human actions using various machine learning approaches. Different models have been developed that apply diverse techniques for detecting and categorizing activities based on image and video data. Features from these datasets are extracted using motion-based and spatial feature learning methods. Many deep learning models have been effectively implemented in this domain to achieve accurate recognition and classification of human activities. These activities include everyday actions such as running, jogging, eating, sitting, and more. HAR has applications across multiple domains, including healthcare, childcare, security, and workplace safety. It plays a significant role in areas such as humancomputer interaction, video surveillance systems, robotics, routine monitoring, and wildlife observation. By utilizing well-known datasets like UCF101, HMDB-51, Hollywood2, and Sports-1M, activity recognition systems can be trained efficiently. The implementation of Convolutional Neural Networks (CNNs), along with tools like OpenCV, enhances the performance of image and video-based recognition models. Applying these datasets enables easier classification of activities based on their nature, whether normal or abnormal. When unusual or suspicious behavior is detected, the system can send real-time alerts to the concerned authorities through a server. Such applications help in preventing harmful activities or at least reducing their impact, making HAR systems highly valuable in ensuring safety and security. Keywords— Human Activity Recognition; Machine Learning;Convolutional Neural Network; OpenCV; video surveillance;