We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
PENERAPAN ALGORITMA CNN (CONVOLUTIONAL NEURAL NETWORK) UNTUK DETEKSI DAN KLASIFIKASI TARGET MILITER BERDASARKAN CITRA SATELIT.
- Authors
Bangun Permadi, Muhamad Luthfi; Gumilang, Restu
- Abstract
The current geopolitical situation in the world tends to be unfavorable, where there is tension in several regional areas, from Europe to the Indo-Pacific. In fact, military confrontation cannot be avoided, it is not impossible that a bigger war will break out, so that Indonesia's readiness is questioned. As a large country, Indonesia's territory is vulnerable to attacks. With weak military strength and technological support, this is certainly a bad thing. In war, the main target is the defense system. Here we try to apply the CNN algorithm to detect and classify military facilities, defense equipment and non-military objects. The goal is that deep learning models can help combat equipment target targets effectively, through satellite imagery. The CNN was tested using several models, and different optimizers, to find out which model had the best accuracy. The model is given image data in three classes, then convolution and max pooling are carried out through its layers, up to 512 neurons in the classification stage and output in 3 neurons according to the number of classes. The result is, the best model using InceptionV3 architecture and Adamax optimizer, produces an accuracy validation value of 96% and the validation loss is 0,1757. As well as a classification report with each class of defense equipment precision of 100%, military facilities of 92%, and non-military of 94%.
- Subjects
INTERNATIONAL relations; ARMED Forces; WEAPONS; MILITARY science; MACHINE learning; REMOTE-sensing images; ACCURACY; MILITARY education
- Publication
Journal of Social & Technology / Jurnal Sosial dan Teknologi (SOSTECH), 2024, Vol 4, Issue 2, p134
- ISSN
2774-5147
- Publication type
Article
- DOI
10.59188/jurnalsostech.v4i2.1138