Advancing Public Activity Recognition in Video Streams Using Hybrid Deep Learning Techniques: A Review

Abstract: Public activity recognition in video streams has become a pivotal area of research due to its applications in surveillance, crowd monitoring, and smart city solutions. Recent advancements in deep learning, particularly hybrid architectures combining multiple learning paradigms, have significantly improved recognition accuracy and robustness in complex environments. This paper reviews the state-of-the-art hybrid deep learning approaches for public activity recognition in video streams, discusses their comparative merits, and outlines open research challenges. The study systematically analyzes methodologies, datasets, and performance metrics used across recent works, providing insights into future research directions.

Keywords: Public Activity Recognition, Video Streams, Hybrid Deep Learning, Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Surveillance.