Detection of diseased pine trees in unmanned aerial vehicle images by using deep convolutional neural networks

Abstract

This study presents a method that uses high-resolution remote sensing images collected by an unmanned aerial vehicle (UAV) and combines MobileNet and Faster R-CNN for detecting diseased pine trees. MobileNet is used to remove backgrounds to reduce the interference of background information. Faster R-CNN is adopted to distinguish between diseased and healthy pine trees. The number of training samples is expanded due to the insufficient number of available UAV images. Experimental results show that the proposed method is better than traditional machine learning approaches, such as support vector machine and AdaBoost, and methods of DCNN, such as Alexnet, Inception and Faster R-CNN. Through sample expansion and background removal, the proposed method achieves effective detection of diseased pine trees in UAV images by using deep learning technology.

Keywords:

• Target detection
• diseased pine tree
• UAV
• sample expansion
• deep convolutional neural networks

Acknowledgements

The authors would like to thank the Shuimuyuan Group for providing some UAV images.

Disclosure statement

The authors declare no conflicts of interest.

Additional information

Funding

This work was supported by the National Natural Science Foundation of China under Grant Nos. 61672032 and 31971789, Major Natural Science Research Projects in Colleges and Universities of Anhui Province under Grant KJ2020ZD03, as well as the Open Research Fund of National Engineering Research Center for Agro-Ecological Big Data Analysis & Application of Anhui University under Grant No. AE201902.