Perancangan Supervisory Control Berbasis Yolov8 Dan Enhanced OCR (EOCR) untuk Operasi Pelabuhan Optimal

Maulidi, Achmad Rafi Issyan (2026) Perancangan Supervisory Control Berbasis Yolov8 Dan Enhanced OCR (EOCR) untuk Operasi Pelabuhan Optimal. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5022221241-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (7MB) | Request a copy

Abstract

Kelancaran operasional pelabuhan sangat bergantung pada proses validasi kendaraan dan kontainer yang cepat, akurat, dan realtime. Namun, proses identifikasi di area gerbang (gate) saat ini masih rentan terhadap human error serta kendala visual, seperti perubahan perspektif, kondisi pencahayaan, dan orientasi teks. Untuk mengatasi masalah tersebut, penelitian ini mengusulkan sistem Supervisory Control berbasis YOLOv8 dan Enhanced Optical Character Recognition (EOCR) guna mengotomatisasi proses validasi. Sistem EOCR yang dikembangkan memadukan DeepLabV3-ResNet50, AAQF, transformasi homografi, image filtering, dan OCR. Data identitas visual yang berhasil diekstrak kemudian diteruskan sebagai masukan bagi Programmable Logic Controller (PLC) berbasis Structured Text (ST) untuk mengeksekusi keputusan operasional di lapangan. Berdasarkan hasil pengujian, model YOLOv8 mencatatkan performa yang sangat baik dengan precision 99,2%, recall 100%, F1-score 99,6%, mAP50 99,3%, dan mAP50-95 sebesar 89,3%. Selain itu, pembacaan informasi kontainer menggunakan EOCR terbukti mampu memvalidasi data kendaraan dan peti kemas secara otomatis. Melalui implementasi Supervisory Control ini, hasil tangkapan computer vision berhasil diterjemahkan dengan baik ke dalam logika kontrol PLC sesuai standar operasional (SOP) yang dirancang. Secara keseluruhan, sistem ini sangat potensial untuk mendukung penerapan automatic gate dan menjadi alternatif solusi modern dalam operasional terminal peti kemas.
==================================================================================================================================
Validating vehicles and containers in time is really important for ports to run smoothly.. When people do this job at port gates they can make mistakes because of things like bad lighting or weird angles. To fix this problem we made a system that uses computers to automate the process. This system uses a tool called YOLOv8 and something called Enhanced Optical Character Recognition to read texts and numbers on containers. The way it works is that it uses a different tools together. One tool helps find the text on the containers another tool figures out the shape of the text. Then it fixes the picture so it is not distorted. After that it uses a kind of computer program to read the text. Once it has all the information it sends it to a controller that makes decisions about what to do. We tested this system. It works really well. The YOLOv8 tool is very good at finding things it is right 99.2% of the time. It never misses anything. We also used tools like DeepLabV3-ResNet50 to help get information from the containers. By using computers to automate this process we can make ports run smoothly and make fewer mistakes. This system is an alternative, for ports that want to use modern technology to manage their containers. The Supervisory Control system we made is very effective. It helps ports run better.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Supervisory Control, YOLOv8, AAQF, DeepLabv3-ResNet50, Structured Text PLC
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7882.P3 Pattern recognition systems T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7888.3 Digital computers
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User:	Achmad Rafi Issyan Maulidi
Date Deposited:	27 Jul 2026 01:13
Last Modified:	27 Jul 2026 01:13
URI:	http://repository.its.ac.id/id/eprint/137806

Actions (login required)

View Item