AUTHOR=Zheng Yueyuan , Wu Gang 

TITLE=Single Shot MultiBox Detector for Urban Plantation Single Tree Detection and Location With High-Resolution Remote Sensing Imagery

JOURNAL=Frontiers in Environmental Science

VOLUME=Volume 9 - 2021

YEAR=2021

URL=https://www.frontiersin.org/journals/environmental-science/articles/10.3389/fenvs.2021.755587

DOI=10.3389/fenvs.2021.755587

ISSN=2296-665X

ABSTRACT=Using high resolution remote sensing images to automatically identify individual trees is of great significance to forestry ecological environment monitoring. Urban forestry has a real need for single tree management such as catkin pollution, maintenance of famous trees, landscape construction and park management. At present, there were problems of missed detection and misdetection in dense plantation and complex background forests. In this study, a single tree detection method based on single shot multi box detector (SSD) is proposed, adjust the feature map, optimize the aspect ratio of the default box, reduce the number of parameters, and get the optimal model of SSD. The optimal SSD model was applied to single tree detection and single tree location detection in campus, orchard and economic plantation, respectively. The experimental results show that the accuracy of single tree detection based on SSD is 96.0%, 92.9%, and 97.6% in Campus basketball court green trees, lychee garden, palm plantation, respectively. It is 11.3% and 37.5% higher than the latest template matching method and CV model method, and 43.1% and 54.2% higher than traditional watershed method and local maximum method. SSD has strong application advantage and potential. This work is of reference value to the application of SSD in other agricultural and forestry products.