Deteksi Kapal di Laut Indonesia Menggunakan Metode You Only Look Once Convolutional Neural Network (YOLO-CNN)

Fandisyah, Adam Fahmi (2020) Deteksi Kapal di Laut Indonesia Menggunakan Metode You Only Look Once Convolutional Neural Network (YOLO-CNN). Undergraduate thesis, Institut TeknoIogi Sepuluh Nopember.

[img] Text
06211640000065-Undergraduate_Thesis.pdf - Accepted Version
Restricted to Repository staff only

Download (3MB) | Request a copy

Abstract

Negara Kesatuan Republik Indonesia (NKRI) adalah negara kepulauan terbesar di dunia. Posisi negara Indonesia yang strategis juga berbatasan langsung dengan 10 negara di laut dan 3 di darat. Bentangan garis pantai Indonesia dengan panjang 81.000 km2, menjadikan laut Indonesia dan wilayah pesisir Indonesia memiliki kandungan kekayaan dan sumber daya alam laut yang sangat berlimpah. Kawasan laut Indonesia masih sering terjadi peristiwa seperti illegal fishing, illegal mining, illegal logging, drugs trafficking dan people smuggling. Kondisi ini menunjukkan bahwa selama ini kurang maksimalnya pengawasan wilayah laut Indonesia. Pesatnya perkembangan teknologi di bidang kecerdasan buatan mendorong ditemukannya deep learning. YOLO-CNN adalah salah satu teknik deep learning yang dikembangkan dengan algoritma untuk mendeteksi sebuah objek secara realtime. Dalam penelitian ini, deteksi tipe kapal dilakukan dengan menggunakan YOLO-CNN dan dievaluasi dengan menghitung nilai Mean Average Precission(mAP) yang dilakukan dengan membandingkan hasilnya dengan ground truth. Hasil deteksi tipe kapal menggunakan YOLO-CNN dengan model yang menggunakan k-means anchor box dapat mengenali tipe kapal pada citra satelit, diperoleh nilai mAP hingga 95,06% pada data training serta 50,41% pada data testing. ======================================================================================================== The Unitary State Republic of Indonesia (NKRI) is the largest archipelagic country in the world. The strategic position of the Indonesian state is also directly adjacent to 10 countries at sea and 3 on land. The stretch of the Indonesian coastline with a length of 81,000 km2, makes the Indonesian sea and Indonesian coastal areas contain abundant marine natural resources. The Indonesian marine area still frequently occurs such as illegal fishing, illegal mining, illegal logging, drugs trafficking and people smuggling. This condition shows that so far the supervision of the Indonesian sea area has not been maximal. The rapid development of technology in the field of artificial intelligence is driving the discovery of deep learning. YOLO-CNN is a deep learning technique developed with algorithms to detect an object in realtime. In this study, vessel type detection was carried out using YOLO-CNN and evaluated by calculating the Mean Average Precission (mAP) value which was carried out by comparing the results with ground truth. The results of ship detection using YOLO-CNN with a model using a k-means anchor box can be accessed from ships on satellite imagery, obtained mAP values of up to 95.06% in training data and 50.41% in data testing.

Item Type: Thesis (Undergraduate)
Uncontrolled Keywords: Convolutionanl Neural Network, Deep Learning, K-means Anchor Box, Mean Average Precision, YOLO Convolutionanl Neural Network, Deep Learning, K-means Anchor Box, Mean Average Precision, YOLO
Subjects: H Social Sciences > HA Statistics
Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques
Divisions: Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User: Adam Fahmi Fandisyah
Date Deposited: 03 Dec 2020 03:42
Last Modified: 03 Dec 2020 03:42
URI: https://repository.its.ac.id/id/eprint/82301

Actions (login required)

View Item View Item