Sistem Klasifikasi Kelompok Varian unit pada Area Night Storage Touch Up Menggunakan Metode Convolutional Neural Network (CNN) Untuk Meningkatkan Efisiensi Waktu

Hibatullah, Ahmad Dzaki (2026) Sistem Klasifikasi Kelompok Varian unit pada Area Night Storage Touch Up Menggunakan Metode Convolutional Neural Network (CNN) Untuk Meningkatkan Efisiensi Waktu. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 2040221134-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

PT Astra Daihatsu Motor (ADM) masih melakukan monitoring jumlah unit pada area Night storage touch up secara manual menggunakan Hand Talkie (HT) dan pencatatan operator. Kondisi tersebut menyebabkan keterlambatan informasi, potensi kesalahan perhitungan, serta risiko penumpukan unit yang dapat mengakibatkan line stop. Penelitian ini bertujuan mengembangkan sistem klasifikasi dan monitoring unit kendaraan secara otomatis berbasis Convolutional Neural Network (CNN) menggunakan algoritma YOLOv8. Penelitian dilakukan melalui pengambilan dan anotasi dataset, pelatihan model YOLOv8s dan YOLOv8m, evaluasi menggunakan Precision, Recall, F1-score, mAP50, mAP50-95, serta implementasi sistem monitoring berbasis Flask secara real time. Sistem dikembangkan untuk mendeteksi dan mengklasifikasikan tiga kategori kendaraan, yaitu Domestic, TMC White, dan TMC Silver. Hasil penelitian menunjukkan bahwa YOLOv8m memberikan performa terbaik dengan nilai Precision 99,8%, Recall 100%, F1-score 99,9%, mAP50 99,5%, dan mAP50-95 95,7%, serta memiliki nilai Box Loss, CLS Loss, dan DFL Loss yang lebih rendah dibandingkan YOLOv8s. Pengujian pada data aktual menghasilkan Precision 99,33%, Recall 100%, F1-score 99,66%, dan Accuracy 99,30%. Berdasarkan hasil tersebut, sistem yang dikembangkan mampu melakukan deteksi dan klasifikasi kendaraan secara real time dengan tingkat akurasi yang tinggi, mempercepat proses monitoring, mengurangi human error, serta mendukung peningkatan efisiensi proses produksi pada area Night storage touch up.
====================================================================================================================================
PT Astra Daihatsu Motor (ADM) currently monitors the number of vehicle units in the Night storage Touch-Up area manually using Handy Talkies (HT) and operator records. This approach leads to information delays, potential counting errors, and the risk of unit accumulation, which may result in production line stoppages. This study aims to develop an automatic vehicle classification and monitoring system based on a Convolutional Neural Network (CNN) using the YOLOv8 algorithm. The research involved dataset acquisition and annotation, training of YOLOv8s and YOLOv8m models, performance evaluation using Precision, Recall, F1-score, mAP50, and mAP50-95, and the implementation of a real time monitoring system based on Flask. The proposed system was designed to detect and classify three vehicle categories: Domestic, TMC White, and TMC Silver. The experimental results show that YOLOv8m outperformed YOLOv8s, achieving a Precision of 99.8%, Recall of 100%, F1-score of 99.9%, mAP50 of 99.5%, and mAP50-95 of 95.7%, while also producing lower Box Loss, Classification (CLS) Loss, and Distribution Focal Loss (DFL) values. Testing under actual production conditions achieved a Precision of 99.33%, Recall of 100%, F1-score of 99.66%, and an Accuracy of 99.30%. These results demonstrate that the proposed system can perform real time vehicle detection and classification with high accuracy, accelerate the monitoring process, reduce human error, and support improved production efficiency in the Night storage touch up area.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	YOLOv8, Night storage touch up, deteksi objek, klasifikasi unit, YOLOv8, night touch up storage, object detection, unit classification
Subjects:	T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques. Image analysis--Data processing.
Divisions:	Faculty of Vocational > 36304-Automation Electronic Engineering
Depositing User:	Ahmad Dzaki Hibatullah
Date Deposited:	05 Aug 2026 01:57
Last Modified:	05 Aug 2026 01:57
URI:	http://repository.its.ac.id/id/eprint/143878

Actions (login required)

View Item