Lenticels are an important structure of the pear,and the illustration of its distribution has an important research significance to reveal the mechanism of pear water loss and optimize the preservation and storage strategy.At present,the analysis of lenticels mainly relies on manual identification,which has the problems of low efficiency and high error,this study adopts the deep learning model YOLOv8 of YOLO series to train the lenticels data set of Korla fragrant pear,and obtains the model of lenticel identification of pear.The performance of yolov8n,yolov8s,yolov8m after training were comprehensively compared.Based on this,the model was evaluated by using the test data set.The experimental results show that yolov8s has both high efficiency and excellent performance in the training of lenticels data set,its recognition accuracy is 96.2%,the recall rate is up to 86.1%,the mAP50 is 0.914,and the F1 score is up to 0.909.The validity of lenticel detection system of the Korla fragrant pear based on the yolov8s was verified,showing the good generalization ability of the model.The use of machine vision to complete the identification of Korla fragrant pear lenticels can improve the efficiency of lenticel analysis and provide technical support for further analysis of the distribution law of lenticels.
维普
万方数据