Detection Method of Coconut Development Intelligence Based on Improved YOLO V5 Model

Because of the special structure of coconut, it is difficult to observe the development of its internal structure accurately, so the research on the development of coconut organs is blank at present, which leads to the quality and quality of coconut can not be guaranteed. To solve this problem, we use CT to carry out non-destructive scanning, to obtain different kinds, different stages of coconut internal image images. In this paper, a novel YOLO V5 model was designed to detect Coconut haustorium and plumule, which are the key structures of Coconut. In order to improve the detection accuracy of coconut haustoriums and buds, an ASFF mechanism and a GAM module were incorporated into YOLO V5 model. The experimental results show that the model can detect the haustoriums and plumules in different periods, and the average precision is improved to 92.43%, which meets the need of high precision detection, thus, it can provide information for intelligent prediction of coconut development for decision making.