Sistem Deteksi Keamanan Pekerja Di Area Translator Menggunakan Metode Convolutional Neural Network (CNN)

Kamajaya, Alif Rangga (2025) Sistem Deteksi Keamanan Pekerja Di Area Translator Menggunakan Metode Convolutional Neural Network (CNN). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 2040211046-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (13MB) | Request a copy

Abstract

Keselamatan kerja merupakan aspek krusial di lingkungan industri, terutama di area dengan aktivitas mesin berat seperti gudang Phonska IV. Area translator menjadi titik rawan karena intensitas aktivitas pemindahan pupuk dari conveyor ke hopper yang berlangsung secara terus-menerus, sehingga berpotensi menimbulkan kecelakaan kerja. Untuk mengatasi hal tersebut, proyek ini mengembangkan sistem deteksi keamanan pekerja berbasis Convolutional Neural Network (CNN) yang terintegrasi dengan alarm otomatis guna memberikan peringatan dini ketika pekerja memasuki zona bahaya. Arsitektur CNN yang digunakan adalah You Only Look Once versi 7 (YOLOv7) sebagai model utama, dan YOLOv8 sebagai pembanding. Sistem ini menganalisis rekaman CCTV untuk mendeteksi keberadaan pekerja secara real-time berdasarkan fitur visual dan pola pergerakan. Pengujian dilakukan pada 20 frame uji dengan tiga kelas: Aman, Bahaya, dan Kosong. Hasil menunjukkan kedua model mencatatkan nilai Precision sebesar 100% untuk kelas Aman dan Bahaya. Namun, YOLOv8 unggul dalam metrik Recall dan F1-Score (100%) untuk kedua kelas, sementara YOLOv7 memperoleh Recall sebesar 85% (Aman) dan 90% (Bahaya), serta F1-Score sebesar 91,89% dan 94,73%. Tingkat akurasi keseluruhan YOLOv7 adalah 91,67%, sedangkan YOLOv8 mencapai 98,34%. Temuan ini menunjukkan bahwa YOLOv8 lebih konsisten dan akurat dalam skenario uji terbatas, serta lebih menjanjikan untuk diimplementasikan dalam sistem deteksi keselamatan kerja berbasis Industrial Internet of Things (IIoT) guna meningkatkan efisiensi dan keberlanjutan operasional perusahaan.
=====================================================================================================================================
Occupational safety is a crucial aspect in industrial environments, especially in areas with heavy machinery activity such as the Phonska IV warehouse. The translator area is a vulnerable point due to the intensity of the continuous activity of transferring fertilizer from the conveyor to the hopper, which has the potential to cause work accidents. To address this, this project developed a worker safety detection system based on a Convolutional Neural Network (CNN) integrated with an automatic alarm to provide early warnings when workers enter a danger zone. The CNN architecture used is You Only Look Once version 7 (YOLOv7) as the main model, and YOLOv8 as a comparison. This system analyzes CCTV footage to detect the presence of workers in real-time based on visual features and movement patterns. Testing was carried out on 20 test frames with three classes: Safe, Danger, and Empty. The results show that both models recorded a Precision value of 100% for the Safe and Danger classes. However, YOLOv8 excelled in Recall and F1-Score (100%) for both classes, while YOLOv7 achieved Recall of 85% (Safe) and 90% (Danger), and F1-Score of 91.89% and 94.73%. The overall accuracy rate of YOLOv7 was 91.67%, while YOLOv8 achieved 98.34%. These findings indicate that YOLOv8 is more consistent and accurate in limited test scenarios, and is more promising for implementation in Industrial Internet of Things (IIoT)-based occupational safety detection systems to improve the efficiency and sustainability of company operations.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Keselamatan kerja, CNN, Deteksi Pekerja, Otomatisasi, IIoT, YOLO; Occupational safety, CNN, Worker Detection, Automation, IIoT, YOLO
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5102.9 Signal processing. T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7882.P3 Pattern recognition systems
Divisions:	Faculty of Vocational > 36304-Automation Electronic Engineering
Depositing User:	Alif Rangga Kamajaya
Date Deposited:	24 Nov 2025 01:31
Last Modified:	24 Nov 2025 01:31
URI:	http://repository.its.ac.id/id/eprint/128812

Actions (login required)

View Item