AIRPLANES DETECTION IN AERIAL IMAGES USING YOLO NEURAL NETWORK

Apply

AIRPLANES DETECTION IN AERIAL IMAGES USING YOLO NEURAL NETWORK

Journal Title: Вісник Національного Авіаційного Університету - Year 2018, Vol 76, Issue 3

Abstract

Purpose: The represented research results are aimed to benchma rk performance of state-of-the-art methods of objects detection. There were tested two popular single-stage neural networks based on the “you only looks once” approach. Methods: convolutional neural network, logi stic regression, probabilistic theory, stochastic gradient descent. Results: The considered artificial neural network architectures for objects detection has been trained and applied for the particul ar task of the airplanes detection in aerial images taken from unmanned aerial vehicles and satellites. Discussion: Presented results of experimental verification prove their high detection ability, lo cation precision and real-time processing speed using modern graphics processing unit. The considered neural networks can be easily re-trained for detection of different classes of ground objects

Authors and Affiliations

Volodymyr Kharchenko, Iurii Chyrka

Keywords

Convolutional neural network; object detection; real-time processing; unmanned aerial vehicle

EP ID EP510264
DOI 10.18372/2306-1472.76.13149
Views 133
Downloads 0