Human Action Recognition Using Pose-Guided Graph Convolutional Networks and Long Short-Term Memory

Human Action Recognition is pivotal in applications like surveillance and human computer interaction. This model introduces a unique approach combining Pose-Guided Graph Convolutional Networks for feature extraction and Long Short-Term Memory networks for both temporal modeling and action classification. The hybrid framework adeptly addresses complex spatial and temporal relationships in human motion data. Pose-Guided Graph Convolutional Network captures features from pose and skeleton data by leveraging inherent body movement structure. Long Short-Term Memory excels in modeling sequences, capturing intricate temporal aspects of actions. Integrating Pose-Guided Graph Convolutional Networks’s spatial insight and Long Short-Term Memory’s classification capabilities, our hybrid model offers a holistic solution for accurate human action recognition and classification. This innovative fusion of Pose-Guided Graph Convolutional Networks and Long Short-Term Memory presents a promising direction, enhancing systems with robustness, accuracy, and comprehensive motion understanding.