Reasearch on Kiwi Fruit Flower Recognition for Efficient Pollination Based on an Improved YOLOv5 Algorithm

A close relationship has been observed between the growth and development of kiwi fruit pollination flower. Flower overlap, flower tilt, other problems will affect this plant’s success rate. model based on YOLOv5 was developed to improve flowers. The K-means++ clustering method used cluster anchors closer target size, which improved speed algorithm. convolutional block module attention mechanism incorporated extraction accuracy with respect features effectively reduce missed detection error rates. optimization function improves recognition overlap tilt angle calculation accurately determines coordinates, point angles. experimental results show that predicted value YOLOv5s is 96.7% its highest. Its mean average precision up 89.1%, F1 score ratio 90.12%, memory requirements are smallest (only 20 MB). achieved highest as determined through a comparison experiment four sets analysed models, thereby demonstrating ability facilitate efficient