Utilizing Webcam Interfaces for Enhanced Human Activity Detection

Abstract:This paper investigates the feasibility and effectiveness of using webcams for real-time human activity recognition (HAR). By leveraging computer vision techniques and deep learning models, specifically Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks, we explore the classification of human tasks through webcam-based video input. The study utilizes the HMDB-51 dataset to train the models and demonstrate their real-time application. Our results show that the proposed approach successfully identifies several human activities in real time, with an accuracy of over 75%, and demonstrates its potential for applications in surveillance, healthcare monitoring, and interactive gaming.