Abstract
Vehicle targets in unmanned aerial vehicle (UAV) images are generally small, so a significant amount of detailed information on targets may be lost after neural computing, which leads to the poor performances of the existing recognition algorithms. Based on convolutional neural networks that utilize the YOLOv3 algorithm, this article focuses on the development of a quick automatic vehicle detection method for UAV images. First, a vehicle dataset for target recognition is constructed. Then, a novel YOLOv3 vehicle detection framework is proposed according to the following characteristics: The vehicle targets in the UAV image are relatively small and dense. The average precision (AP) increased by 5.48%, from 92.01% to 97.49%, which still remains the rather high processing speed of the YOLO network. Finally, the proposed framework is tested using three datasets: COWC, VEDAI, and CAR. The experimental results demonstrate that our method had a better detection capability.
Highlights
Vehicle detection in unmanned aerial vehicle (UAV) images is valuable for both the civil and military applications
This paper proposes a fast vehicle detection framework based on a novel convolutional neural network, YOLOv3 [18]
In this work, according to the characteristics of vehicles in UAV images, we applied the K-means++ algorithm [19] to improve the recognition performance of the YOLOv3 network, and used the Soft-NMS [20] to relieve the problem of the wrong multi-box
Summary
Vehicle detection in unmanned aerial vehicle (UAV) images is valuable for both the civil and military applications. Numerous studies have been conducted using neural networks to improve vehicle detection performance from UAV images. The core of machine learning (ML) [3,4] is to learn from data. ML methods include a support vector machine (SVM) [5,6], artificial neural network, naive Bayes, random forest, logistic regression, and adaptive boosting (AdaBoost) in engineering practice [7]. The accuracy of pedestrian detection based on SVM is greatly improved, and Luo et al [8] reported a novel method for a facial expression recognition algorithm that employs core local binary pattern (LBP) information
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
