AMANULLAH, ADITYA DIFA (2025) PEMANFAATAN YOLOV11 DALAM DETEKSI OBJEK TEMPUR BERBASIS CITRA DRONE DI MEDAN PERTEMPURAN. Skripsi thesis, UPN Veteran Yogyakarta.
![[thumbnail of 123210085_Abstrak.pdf]](http://eprints.upnyk.ac.id/style/images/fileicons/text.png) |
Text
123210085_Abstrak.pdf Download (232kB) |
|
Text
123210085_Cover.pdf Download (159kB) |
|
Text
123210085_Daftar Isi.pdf Download (324kB) |
|
Text
123210085_Daftar Pustaka.pdf Download (190kB) |
|
Text
123210085_Halaman Pengesahan Pembimbing.pdf Download (677kB) |
|
Text
123210085_Halaman Pengesahan Penguji.pdf Download (554kB) |
|
Text
123210085_Nashkah Tugas Akhir.pdf Restricted to Repository staff only Download (2MB) |
Abstract
ABSTRAK
Sepanjang sejarah, perang terus mendorong perkembangan teknologi
militer, termasuk drone yang sejak 2010 digunakan untuk pengintaian dan serangan.
Dalam perang Ukraina–Rusia, drone komersial lebih dipilih karena murah, sulit
terdeteksi, dan efektif untuk misi pengintaian, koreksi artileri, hingga serangan
kamikaze, sekaligus meningkatkan keamanan personel. Kini drone dipadukan
dengan kecerdasan buatan untuk melampaui kinerja tim pengintaian, meski drone
komersial masih bergantung pada pemrosesan eksternal karena tidak dilengkapi AI
bawaan. Penelitian ini menggunakan algoritma YOLOv11s, yang merupakan versi
terbaru dari YOLO. Penelitian ini bertujuan untuk mengetahui performa YOLOv11
dalam mendeteksi objek tempur berbasis citra drone di medan pertempuran.
Dataset didapatkan dari Roboflow Universe dan scrapping dari video
Youtube yang terdiri dari 7.594 gambar dengan detail yaitu 2.753 untuk kendaraan
lapis baja ringan(light), 5.228 untuk kendaraan lapis baja berat(tank) dan 387
terdapat gambar kendaraan lapis baja ringan dan tank. Dataset dilatih menggunakan
model YOLOv11s sebanyak 200 epoch kemudian hasil pelatihan digunakan pada
sistem.
Hasil pengujian terhadap data validasi menunjukan bahwa model
YOLOv11s mencapai nilai presisi sebesar 0.7484, recall sebesar 0.6911, dan F1
score sebesar 0.7185 untuk class light sedangkan untuk class tank 0.8365 untuk
presisi, 0.7824 untuk recall, dan 0.8084 untuk F1-score. Model mampu mendeteksi
objek tempur berbasis citra drone meskipun terdapat beberapa kesalahan seperti
salah deteksi pada objek yang disebabkan bentuk yang mirip atau warna yang sama
dengan background.
Kata Kunci : YOLOv11, Deep Learning, Deteksi Objek Tempur
vii
ABSTRACT
Throughout history, war has continuously driven the development of
military technology, including drones, which have been used for reconnaissance
and attack since 2010. In the Ukraine-Russia war, commercial drones were
preferred because they were inexpensive, difficult to detect, and effective for
reconnaissance missions, artillery corrections, and kamikaze attacks, while also
improving personnel safety. Now, drones are being combined with artificial
intelligence to surpass the performance of reconnaissance teams, although
commercial drones still rely on external processing because they lack built-in AI.
This research used the YOLOv11s algorithm, the latest version of YOLO. This study
aimed to determine the performance of YOLOv11 in detecting combat objects based
on drone imagery on the battlefield.
The dataset was obtained from Roboflow Universe and scraped from
YouTube videos. It consisted of 7,594 images with details: 2,753 for light armored
vehicles, 5,228 for heavy armored vehicles (tanks), and 387 for light armored
vehicles and tanks. The dataset was trained using the YOLOv11s model for 200
epochs, and the training results were then used in the system.
Test results on validation data show that the YOLOv11s model achieved a
precision of 0.7484, a recall of 0.6911, and an F1-score of 0.7185 for the light class,
while for the tank class, the precision was 0.8365, the recall was 0.7824, and the
F1-score was 0.8084. The model was able to detect combat objects based on drone
imagery, although it encountered several errors, such as false detections of objects
due to similar shapes or colors to the background.
Keywords: YOLOv11, Deep Learning, Combat Object Detection
viii
| Item Type: | Tugas Akhir (Skripsi) |
|---|---|
| Uncontrolled Keywords: | YOLOv11, Deep Learning, Combat Object Detection |
| Subjek: | Z Bibliography. Library Science. Information Resources > ZA Information resources |
| Divisions: | Fakultas Teknik Industri > (S1) Informatika |
| Depositing User: | Eko Yuli |
| Date Deposited: | 28 Oct 2025 06:41 |
| Last Modified: | 28 Oct 2025 06:41 |
| URI: | http://eprints.upnyk.ac.id/id/eprint/45099 |
Actions (login required)
 |
View Item |