Joint-level voting and action clustering for enhanced skeleton-based human action recognition

Puntos clave

• Enhanced human action recognition is achieved through joint-level voting and action clustering techniques.
• The analysis used joint-level voting, improving accuracy in recognizing complex movements like dancing and sports.
• Evaluation on various action datasets highlights the effectiveness of the proposed framework in action classification.
• Success in human action recognition may enable applications in security, entertainment, and robotics.