Low Altitude and Low Speed Uav Identification based on Hybrid Model

DOI

10.2991/iccia-19.2019.12How to use a DOI?

Keywords

Uav identification, Motion blur, Image denoising.

Abstract

The identification of low-altitude and low-speed uav (Unmanned aerial vehicle) is a hot issue in the field of computer vision. Most problems can be solved based on traditional identification methods, but there is a blind spot for low-altitude uav. By contrast, the method based on deep learning can better solve this problem, but in the case of noise and motion blur, the processing effect of CNN and other methods is poor. In order to solve this problem, we put forward a new model. On the basis of the original residual model, we adopt the convolution kernel of different sizes and combine Inception block with multi-scale convolution group. As the low-altitude uav still has the speed, it needs multi-scale convolution to expand the accepted features for identification. Specifically, the residual connection can accelerate our training and meet the real-time requirements. The enlarged convolution kernel can accept more features to satisfy our identification in the case of noise, and multiple ways of convolving kernel concatenate can satisfy our identification in the case of motion blur in the flying uav. Therefore, this paper aims to train a uav with multi-scale convolution kernel, which can effectively identify low-altitude and slow-flying uav under the condition of noise and motion blur. Experimental results show that this method is feasible.

Copyright

© 2019, the Authors. Published by Atlantis Press.

Open Access

This is an open access article distributed under the CC BY-NC license (http://creativecommons.org/licenses/by-nc/4.0/).

Download article (PDF)

TY  - CONF
AU  - Kai Luo
AU  - Guibin Zhu
AU  - Ying Li
AU  - Wentao Wang
PY  - 2019/07
DA  - 2019/07
TI  - Low Altitude and Low Speed Uav Identification based on Hybrid Model
BT  - Proceedings of the 3rd International Conference on Computer Engineering, Information Science & Application Technology (ICCIA 2019)
PB  - Atlantis Press
SP  - 76
EP  - 87
SN  - 2352-538X
UR  - https://doi.org/10.2991/iccia-19.2019.12
DO  - 10.2991/iccia-19.2019.12
ID  - Luo2019/07
ER  -