Grading Of Auricularia Auriculata Based On Machine Vision And Deep Learning

Classification and detection belongs to the category of classification recognition.After the object to be detected is collected,computer technology is used to extract the feature information of the image,and according to the extracted effective information to complete the classification of different levels.The research of machine vision and deep learning has been applied widely for the classification and detection tasks in the field of agricultural products.Machine vision is a method that can realize the extraction of color,shape parameters,texture and other information,which can complete the task of classification and detection with high efficiency and accuracy.The reason why deep learning research is appropriate in the tasks of large-scale image classification and detection is that multi-scale feature information can be extracted by multi-level convolutional neural network that through image convolution operation and sampling operation.(1)In this paper,the image of Auricularia nigricans was taken under different backgrounds,and the contrast ratio of the foreground and background area of the image was analyzed,so as to determine the best background for the image acquisition of Auricularia nigricans.(2)After the author collected the image of Auricularia nigricans,through the contrast analysis of machine vision in the color space model method to eliminate Auricularia nigricans with mildew,the R component of RGB model is chosen as the final judging basis.After statistical analysis of the distribution range of R values,the author took the optimal threshold value as the discriminant standard to eliminate mildewed Auricularia nigricans and ensure food safety.(3)After the mildewed Auricularia nigricans was removed,the image of marked Auricularia nigricans was divided into a data set and a test set in a ratio of 8:2,and the training test was conducted in multiple convolutional neural network models built respectively.After the comparative analysis of the experimental results,the convolutional neural network model with the highest test accuracy was finally selected to realize the classification and detection of Auricularia nigricans.In this paper,the R component in the RGB model was used as the threshold standard to remove the mildewed Auricularia nigricans,and the recognition accuracy reached 95%.Then,in the deep learning classification and detection,the RESNET101 network model with the highest test accuracy was selected by comparing the experimental results,and the classification and detection accuracy rate of 94.32% was achieved.