Model Simulasi Deteksi dan Sistem Peringatan Dini Penggunaan APD pada Proyek Konstruksi Berbasis Hybrid YOLO–Fuzzy

Al Hakim, Muhammad Rais Fahd (2026) Model Simulasi Deteksi dan Sistem Peringatan Dini Penggunaan APD pada Proyek Konstruksi Berbasis Hybrid YOLO–Fuzzy. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6012241065-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Keselamatan kerja merupakan aspek krusial dalam industri konstruksi yang memiliki tingkat risiko kecelakaan tertinggi dibanding sektor lainnya. Ketidakpatuhan pekerja terhadap penggunaan Alat Pelindung Diri (APD) menjadi salah satu penyebab dominan kecelakaan, sehingga pengawasan APD perlu dilakukan secara konsisten. Penelitian ini bertujuan untuk merancang model deteksi APD menggunakan algoritma YOLOv11, mengembangkan model pengambilan keputusan berbasis logika fuzzy Mamdani, serta mengintegrasikan dan menguji keduanya sebagai sistem Hybrid YOLO-Fuzzy Inference System (HYFIS) pada aktivitas pekerjaan konstruksi. Model deteksi APD dirancang menggunakan YOLOv11m untuk mengenali lima jenis APD (helm, rompi, sarung tangan, sepatu, dan sabuk pengaman) pada video aktivitas pekerja, dan mencapai tingkat ketepatan deteksi sebesar 83,8% dengan pengenalan di atas 80% pada seluruh jenis APD. Hasil deteksi kemudian dikonversi menjadi skor kepatuhan sesuai bobot kebutuhan APD tiap jenis pekerjaan dan diolah melalui logika fuzzy Mamdani untuk menghasilkan empat kategori respons keselamatan, yaitu Aman, Teguran, Alarm Lokal, dan Hentikan Kerja. Pengujian menunjukkan keempat kategori dihasilkan secara konsisten dengan peralihan antartingkat yang halus. Kedua model diintegrasikan menjadi HYFIS dan diuji pada pekerjaan pasangan bata, pengecoran, dan plaster. Hasil pengujian menunjukkan 13 dari 15 video uji (86,7%) menghasilkan respons keselamatan yang sesuai dengan penilaian manual. Model yang dikembangkan mampu mendeteksi kepatuhan APD dan memberikan peringatan dini secara otomatis, sehingga dapat mendukung pengawasan keselamatan kerja dan penerapan budaya K3 berbasis teknologi di sektor konstruksi.
===================================================================================================================================
Occupational safety is a crucial aspect of the construction industry, which has the highest accident risk compared to other sectors. Workers' non-compliance with Personal Protective Equipment (PPE) use remains a dominant cause of accidents, necessitating consistent PPE monitoring. This study aims to develop a PPE detection model using the YOLOv11 algorithm, design a decision-making model based on Mamdani fuzzy logic, and integrate and test both as a Hybrid YOLO-Fuzzy Inference System (HYFIS) for construction work activities. The PPE detection model was developed using YOLOv11m to recognize five types of PPE (helmets, vests, gloves, safety boots, and safety harnesses) in worker activity videos, achieving a detection accuracy of 83.8% with recognition rates above 80% across all PPE types. Detection results were then converted into compliance scores based on PPE requirement weights for each job type and processed through Mamdani fuzzy logic to produce four safety response categories: Safe, Warning, Local Alarm, and Stop Work. Testing confirmed that all four categories were generated consistently with smooth transitions between levels. Both models were integrated into HYFIS and tested on bricklaying, concrete, and plastering work. Results showed that 13 out of 15 test videos (86.7%) produced safety responses consistent with manual assessment. The developed model is capable of detecting PPE compliance and providing automated early warnings, thereby supporting workplace safety monitoring and the implementation of technology-driven safety culture in the construction sector.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	keselamatan kerja, alat pelindung diri, deteksi objek, yolo, logika fuzzy, konstruksi, occupational safety, personal protective equipment, object detection, yolo, fuzzy logic, construction
Subjects:	T Technology > T Technology (General) > T55 Industrial Safety T Technology > T Technology (General) > T58.62 Decision support systems T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques. Image analysis--Data processing.
Divisions:	Faculty of Civil, Planning, and Geo Engineering (CIVPLAN) > Civil Engineering > 22101-(S2) Master Thesis
Depositing User:	Muhammad Rais Fahd Al Hakim
Date Deposited:	27 Jul 2026 01:27
Last Modified:	27 Jul 2026 01:27
URI:	http://repository.its.ac.id/id/eprint/137364

Actions (login required)

View Item