Sistem Deteksi Postur Duduk Ergonomis Menggunakan Pose Estimation dan Support Vector Machine

Priyambodo, Danar Sodik (2025) Sistem Deteksi Postur Duduk Ergonomis Menggunakan Pose Estimation dan Support Vector Machine. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5025211145-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (7MB) | Request a copy

Abstract

Musculoskeletal Disorders (MSD) merupakan salah satu permasalahan kesehatan yang umum terjadi di lingkungan kerja akibat postur duduk yang tidak ergonomis. Dalam penelitian ini, dikembangkan sebuah sistem deteksi postur kerja secara otomatis berbasis pose estimation dan Support Vector Machine (SVM). Sistem memanfaatkan keypoints hasil estimasi dari OpenPose untuk menghitung sudut tubuh utama seperti bahu, siku, leher, punggung, pergelangan tangan, dan lutut. Nilai sudut tersebut kemudian digunakan sebagai fitur untuk klasifikasi tingkat risiko ergonomis berdasarkan metode RULA dan REBA. Sistem dikembangkan dengan tiga jenis input, yaitu prediksi dari gambar, video, dan prediksi real-time menggunakan websocket. Evaluasi performa menunjukkan bahwa model SVM-RULA menghasilkan performa terbaik dengan akurasi sebesar 96%, precision sebesar 90%, recall sebesar 94%, dan F1-score sebesar 92% pada data uji. Validasi menggunakan 10 gambar nyata menunjukkan tingkat kesesuaian sebesar 80% terhadap penilaian ahli. Hasil menunjukkan bahwa sistem mampu mendeteksi postur kerja secara objektif dan memberikan feedback secara langsung, sehingga dapat menjadi solusi untuk peringatan dini terhadap postur tidak ergonomis dan mendukung peningkatan kesehatan pekerja sesuai dengan tujuan Sustainable Development Goals (SDG) Kesehatan dan Produktivitas.
=================================================================================================================================
Musculoskeletal Disorders (MSDs) are a common health issue in the workplace, often caused by non-ergonomic sitting postures. This study developed an automatic system for detecting work posture based on pose estimation and Support Vector Machine (SVM). The system utilizes keypoints estimated by OpenPose to calculate the main body joint angles, such as shoulders, elbows, neck, trunk, wrists, and knees. These angle values are then used as features to classify the level of ergonomic risk based on the RULA and REBA methods. The system supports three types of input: image-based prediction, video-based prediction, and real-time prediction using websocket. Performance evaluation shows that the SVM-RULA model achieved the best results, with an accuracy of 96%, precision of 90%, recall of 94%, and an F1-score of 92% on the test data. Validation using 10 real posture images showed an 80% agreement rate with expert assessment. These results indicate that the system can objectively detect work posture and provide real-time feedback, making it a promising solution for early warnings of non-ergonomic postures and supporting the improvement of worker health in line with the Sustainable Development Goals (SDG) on Health and Productivity.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Ergonomi, Kesehatan, Musculoskeletal Disorders, Pose Estimation, Produktivitas, Support Vector Machine, Ergonomics, Health, Musculoskeletal Disorders, Pose Estimation, Productivity, Support Vector Machine
Subjects:	T Technology > T Technology (General) > T174.5 Technology--Risk assessment. T Technology > T Technology (General) > T385 Visualization--Technique T Technology > T Technology (General) > T57.5 Data Processing T Technology > T Technology (General) > T57.8 Nonlinear programming. Support vector machine. Wavelets. Hidden Markov models. T Technology > T Technology (General) > T58.62 Decision support systems
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Danar Sodik Priyambodo
Date Deposited:	24 Jul 2025 08:43
Last Modified:	24 Jul 2025 08:43
URI:	http://repository.its.ac.id/id/eprint/121316

Actions (login required)

View Item