Klasifikasi Multi-Label Postur Tubuh Berbasis Mediapipe dengan Perbandingan Model Machine Learning

Fridy Mandita; Abdul Azizur Rokhman

doi:10.22441/format.2026.v15.i2.007

Authors

Fridy Mandita Universitas 17 Agustus 1945 Surabaya, Indonesia
Abdul Azizur Rokhman Universitas 17 Agustus 1945 Surabaya, Indonesia

DOI:

https://doi.org/10.22441/format.2026.v15.i2.007

Abstract

Penelitian ini mengembangkan sistem klasifikasi multi-label untuk mengenali indikasi Forward Head Posture (FHP), Postural Kyphosis (PK), Rounded Shoulder Posture (RSP), dan Normal Posture (NP) melalui analisis citra postur statis berdiri yang dapat merepresentasikan perubahan postural berkaitan dengan gaya hidup sedentari. Meskipun demikian, pengukuran dilakukan melalui analisis citra postur statis dalam posisi berdiri untuk mengamati keselarasan kepala, bahu, tulang belakang, dan panggul. MediaPipe Pose digunakan untuk mengekstraksi 33 landmark tubuh dari setiap citra. Dataset awal terdiri atas 933 citra yang diperoleh dari dataset publik POLAR dan hasil kurasi mandiri. Setelah pemeriksaan kualitas citra dan kelengkapan landmark, sebanyak 54 citra dikeluarkan sehingga diperoleh 879 citra valid. Data dibagi menggunakan Iterative Stratified Split dengan rasio 80:20 untuk mempertahankan distribusi kombinasi label. Sebanyak 132 atribut mentah hasil ekstraksi landmark dan 54 fitur hasil feature engineering membentuk 186 fitur awal. Seleksi fitur menggunakan Mutual Information pada data latih menghasilkan 60 fitur terpilih. Model yang dibandingkan meliputi Light Gradient Boosting Machine, Support Vector Machine, dan Artificial Neural Network. Hasil pengujian menunjukkan bahwa SVM memberikan performa terbaik dengan F1-score macro sebesar 0,76, F1-score micro sebesar 0,79, dan Hamming Loss sebesar 0,19. FHP menjadi kelas dengan performa tertinggi, sedangkan PK menjadi kelas yang paling sulit dikenali karena keterbatasan landmark dua dimensi pada area vertebra torakal. Hasil penelitian menunjukkan bahwa kombinasi MediaPipe Pose, feature engineering, dan machine learning berpotensi mendukung pengembangan sistem pemantauan indikasi postur secara waktu nyata.

Downloads

Download data is not yet available.

References

[1] F. Roggio, S. Di Grande, S. Cavalieri, D. Falla, and G. Musumeci, “Biomechanical Posture Analysis in Healthy Adults with Machine Learning: Applicability and Reliability,” Sensors, vol. 24, no. 9, May 2024, doi: 10.3390/s24092929.

[2] J. E. Estrada, L. A. Vea, and M. Devaraj, “Modelling Proper and Improper Sitting Posture of Computer Users Using Machine Vision for a Human–Computer Intelligent Interactive System during COVID-19,” Applied Sciences (Switzerland), vol. 13, no. 9, May 2023, doi: 10.3390/app13095402.

[3] G. Çobanoğlu, M. Y. Demirkan, Z. B. Ecemiş, and N. Atalay Güzel, “Forward Head Posture and Its Effect on Muscle Activation,” Gazi Sağlık Bilimleri Dergisi, vol. 9, no. 1, pp. 85–93, Mar. 2024, doi: 10.52881/gsbdergi.1376080.

[4] S. Tavakoli and A. Maleki, “Proposing Myoelectric Biomarkers for Assessing Kyphosis,” InfoScience Trends, vol. 2, no. 8, pp. 44–56, Aug. 2025, doi: 10.61882/ist.202502.08.04.

[5] M. Khan et al., “RELATIONSHIP OF ROUNDED SHOULDER POSTURE WITH PEAK EXPIRATORY FLOW RATE IN UNDERGRADUATE MEDICAL STUDENTS HAVING MECHANICAL NECK PAIN,” 2024.

[6] W. Abu-Taleb, A. A. Yamany, Y. M. Aneis, and S. T. Abu El Kasem, “Effect of adding global postural reeducation to kendall exercises for treating asymptomatic forward head posture: A single-blinded randomized controlled trial,” J. Bodyw. Mov. Ther., vol. 42, pp. 938–947, Jun. 2025, doi: 10.1016/j.jbmt.2025.01.039.

[7] F. Roggio, B. Trovato, M. Sortino, and G. Musumeci, “A comprehensive analysis of the machine learning pose estimation models used in human movement and posture analyses: A narrative review,” Nov. 15, 2024, Elsevier Ltd. doi: 10.1016/j.heliyon.2024.e39977.

[8] M. Menanno, C. Riccio, V. Benedetto, F. Gissi, M. M. Savino, and L. Troiano, “An Ergonomic Risk Assessment System Based on 3D Human Pose Estimation and Collaborative Robot,” Applied Sciences (Switzerland), vol. 14, no. 11, Jun. 2024, doi: 10.3390/app14114823.

[9] I. Nurdiansyah, R. Utami, and ; Muchamad Sandy, “Machine Learning System untuk Mendeteksi Gerakan Tubuh Menggunakan Library Mediapipe,” Jan. 2025. doi: https://doi.org/10.22441/format.2025.v14.i1.008.

[10] P. Kongchansawang, T. Naowavathong, R. Srisuttee, and S. Moodleah, “Automated Classification of User Exercise Poses in Virtual Reality Using Machine Learning-Based Human Pose Estimation,” IEEE Access, vol. 13, pp. 179789–179802, 2025, doi: 10.1109/ACCESS.2025.3622115.

[11] J. H. Meghana, H. K. Chethan, K. S. Santhosh Kumar, and S. P. S. Prakash, “Comprehensive Analysis of Pose Estimation and Machine Learning Classifiers for Precise Yoga Pose Detection and Classification,” in Procedia Computer Science, Elsevier B.V., 2025, pp. 3345–3356. doi: 10.1016/j.procs.2025.04.592.

[12] S. Takigami et al., “Estimation of Shoulder Joint Rotation Angle Using Tablet Device and Pose Estimation Artificial Intelligence Model,” Sensors, vol. 24, no. 9, May 2024, doi: 10.3390/s24092912.

[13] J. Shen, L. Chen, X. He, C. Zuo, X. Li, and L. Dong, “An Interactive Human-in-the-Loop Framework for Skeleton-Based Posture Recognition in Model Education,” Biomimetics, vol. 10, no. 7, Jul. 2025, doi: 10.3390/biomimetics10070431.

[14] H. Zhang, D. Gračanin, W. Zhou, D. Dudash, and G. Rushton, “Toward Real-Time Posture Classification: Reality Check,” Electronics (Switzerland), vol. 14, no. 9, May 2025, doi: 10.3390/electronics14091876.

[15] A. Tharatipyakul, T. Srikaewsiew, and S. Pongnumkul, “Deep learning-based human body pose estimation in providing feedback for physical movement: A review,” Sep. 15, 2024, Elsevier Ltd. doi: 10.1016/j.heliyon.2024.e36589.

[16] A. Peryanto, A. Yudhana, and D. R. Umar, “Rancang Bangun Klasifikasi Citra Dengan Teknologi Deep Learning Berbasis Metode Convolutional Neural Network,” 2019. [Online]. Available: https://www.mathworks.com/discovery/convolutional-neural-network.html

[17] Y. Liu et al., “Representation-centric survey of supervised skeletal action recognition and the new benchmark,” Pattern Recognit., vol. 180, p. 114140, Dec. 2026, doi: 10.1016/j.patcog.2026.114140.

[18] M. R. Abrori, F. T. Anggraeny, and E. Y. Puspaningrum, “Analisis Fitur Pada Citra Gestur Tangan Sistem Isyarat Bahasa Indonesia,” FORMAT: Jurnal IlmiahTeknik Informatika, vol. 12, no. 2, pp. 165–172, Dec. 2023.

[19] I. H. Al Ghozali, M. A. Fathin, and A. R. Handoko, “A Comparative Study of Machine Learning with Statistical Feature Selection for Risk Detection of Diabetic,” Jurnal Ilmiah FIFO, vol. 17, no. 2, p. 102, Nov. 2025, doi: 10.22441/fifo.2025.v17i2.001.

[20] T. H. Pinem and Z. P. Putra, “Evaluasi Kinerja Algoritma Klasifikasi Deep Learning dalam Prediksi Diabetes,” Jurnal Ilmiah FIFO, vol. 17, no. 1, p. 17, Jun. 2025, doi: 10.22441/fifo.2025.v17i1.003.