Graph-Based Architecture for Consumer-Based Image and Video Analysis in the Internet of Everything Using Advanced Learning Algorithms

The Internet of Everything (IoE) era has witnessed to a rapid increase in interconnected devices, leading to an exponential growth of image and video data. Efficient multimedia content analysis is crucial for applications such as surveillance systems and autonomous vehicles. This study proposes a novel hybrid architecture, fusing Convolutional Neural Networks (CNNs) with Graph Convolutional Networks (GCNs), for efficient multimedia analysis within the burgeoning Internet of Everything (IoE) landscape. By synergistically leveraging the strengths of both CNNs for local feature extraction and GCNs for capturing global data relationships, this approach aims to significantly enhance the accuracy and robustness of image and video analysis tasks in diverse IoE applications. Evaluated on challenging datasets like YouTube-8M and ImageNet, our approach demonstrates significant improvements in accuracy, loss, and computational efficiency compared to traditional methods. These results highlight the potential of this hybrid framework to address the demands of the IoE by enabling more efficient and accurate analysis of the ever-growing volume of image and video data.