A Methodfor Detectionand Identification of Fritillaria Cirrhosa D.Don Based on Improved YOLOv3
Since the authentic Fritillaria Cirrhosa D.Don resources are scare due to its high valuable medical uses often occurs in the market,which seriously affects the quality of Fritillaria.At present,the identification of Fritillaria mainly relies on traditional trait identification,microscopic identification,physical and chemical identification and etc.,which is subjective and requires high practical experience for operators,and the preprocessing is cumbersome.In view of the shortcomings of the current identification methods of Fritillaria,adopt deep learning method to realize the automatic detection and recognition of Fritillaria,and improve YOLOv3 to train different classes of image data,embeds dual-path modules and channel attention mechanism,respectively strengthen the models'expression of the features of Fritillaria from the feature extraction and feature selection.Experiments results show that this model realize the rapid batch automatic identification of Fritillaria,and the recognition mean average precision can reach 80％,providing a new solution for the quality evaluation of Fritillaria in the Chinese herbal medicine industry.
万方数据
维普
