Strawberry Maturity Classification from UAV and Near-Ground Imaging Using Deep Learning

• DOI: 10.1016/j.atech.2021.100001
• Publication Date: 2021-09-16T10:10:31Z
• Authors: Xue Zhou; Won Suk Lee; Yiannis Ampatzidis; Yang Chen; Natalia Peres; Clyde Fraisse
• Keywords: Digital camera; Machine Vision

Strawberry is ranked third in the value of production crops Florida, USA. Classifying strawberry maturity and monitoring growth status field very critical for accurate yield prediction, efficient management, achieving highest crop production. The traditional method distinguishing based on either physical appearance or internal chemical substance content. However, time-consuming costly. In this research, an automatic classification system was developed rapid different stages. A state-of-the-art deep learning method, You Only Look Once (YOLOv3), which good at small object detection, trained applied to detect flowers fruit Two image acquisition methods, aerial imaging near-ground imaging, were compared by using same processing method. As a result, three seven stages classified unmanned vehicle (UAV) images digital camera images, respectively. For UAV mean average precision (mAP) 0.88 test data set 2 m, (AP) 0.93 fully matured fruit. mAP 0.89, AP 0.94 as well. result shows that YOLOv3 excellent approach both types.