Abstract: Bharatanatyam is a classical Indian dance form which involves rigorous structuring and exacting coordination of posture, speed, balance, and emotions through physical expressions. In conventional performance evaluation, expert observations play a major role in evaluation process, thus resulting in subjectivity and inconsistency in evaluation process. This research project proposes a comprehensive computational approach for objective biomechanical evaluation and automated analysis of Bharatanatyam performances using computer vision, biomechanics and artificial intelligence. Markerless pose estimation technique was used to estimate skeletal landmarks from Bharatanatyam performance videos. These landmark points were preprocessed by performing operations like interpolation, normalization of coordinates, temporal resampling and signal filtering in order to produce accurate motion trajectories. Afterward, kinetic parameters such as joint angles, velocities, acceleration and mechanical energy were computed to analyze the dynamics of movement. A variety of machine learning models, including but not limited to Random Forest, LSTM, Isolation Forest, and Simulated Annealing have been used for the purpose of movement classification, temporal pattern analysis, anomaly detection, and biomechanics optimization, correspondingly. The analysis of the kinematics showed significant differences between the joint mechanics and mechanical energy of different types of adavus, proving the biomechanics of Bharatanatyam dance. The most important parameters for the dance motion classification using the Random Forest model were acceleration, rhythm stability, posture stability, and joint angle features, while the LSTM neural network had the ability to detect temporal dependencies within the sequence of dance motions. Moreover, the optimization algorithms made the dance motion more efficient, decreasing its mechanical energy without changing the structure of the dance motion.
Keywords: Bharatanatyam; Biomechanics; Pose estimation: Kinematics Aanalysis: Machine Learning.
Title: A COMPUTATIONAL FRAMEWORK FOR BIOMECHANICAL ANALYSIS AND ARTIFICIAL INTELLIGENCE-BASED ASSESSMENT OF BHARATANATYAM MOVEMENTS
Author: Anju Devi Kisson
International Journal of Interdisciplinary Research and Innovations
ISSN 2348-1218 (print), ISSN 2348-1226 (online)
Vol. 14, Issue 3, July 2026 - September 2026
Page No: 128-136
Research Publish Journals
Website: www.researchpublish.com
Published Date: 23-September-2026
