Research On Chinese Herbal Medicine Image Classification Method Based On Deep Learning

Chinese herbal medicine has good effects in treating some diseases and has made great contributions to human health.But there are many kinds of Chinese herbal medicine.The same kind of Chinese herbal medicine will appear different forms due to the influence of growth environment and other factors.Some different kinds of Chinese herbal medicine plants are very similar,so it is a difficult task to classify them accurately.However,there are few studies on the classification of Chinese herbal medicines,and the existing studies only classify a few plants.This paper designs a classification method based on the image of Chinese herbal medicine to realize the accurate classification of Chinese herbal medicine.At present,there is not a very comprehensive database for Chinese herbal medicine images,so we first collected and established a large amount of Chinese herbal medicine image data sets.In recent years,deep learning has developed rapidly,and it has achieved very good results in classification and recognition.Therefore,we use the deep learning method to classify the collected images of Chinese herbal plants.We first classify the original Chinese herbal image dataset.By analyzing the experimental results,we find that the background of the image has a great influence on the classification results.Because most of the images are taken in a natural state,it is inevitable that there will be a very complicated background,which causes considerable interference in the classification.In order to eliminate the complex background problems in the Chinese herbal medicine image,we segmented the image.Image segmentation uses a combination of automatic image segmentation and manual image segmentation to improve image segmentation.Firstly,a part of the segmented image is obtained by the image segmentation algorithm.After that,the images with poor segmentation effect are manually segmented;these segmented images are sorted out as data sets;finally,they are classified by deep learning algorithm,in the end,the accuracy has been improved correspondingly,and the accuracy of TOP-1 is 72%,and the accuracy of TOP-5 is 93.6%...