Ingredients Image Classification Algorithm Research And Design Based On Enhanced Feature CNN

The catering industry is one of the country's major industries.Receipt of ingredients is an important part of controlling catering safety and quality.How to quickly identify and classify ingredients is the core and key issue of automatic food receipt.Due to the variety of ingredients and prices,and the quality of the images captured during the automated receipt process is difficult to control,these differences pose a great challenge to the fast and accurate classification of food images.Therefore,how to quickly and accurately classify food images is one of the difficult problems to be solved when the quality of food images is uneven.Convolutional neural network(CNN)is a deep neural network commonly used in the field of computer vision.AlexNet,VGGNet,and ResNet are all using CNN to obtain the best classification effect in the field of image recognition.The CNN model extracts all the features of the image from the local to the whole through cascaded convolutional structure.Compared with the traditional method,the CNN model has great advantages in the accuracy and time of image classification and has a profound impact on the development of the computer vision field.In this paper,under the environment of restaurant ingredients receiving,we proposed a food ingredient classification algorithm based on enhanced feature(order characteristics and weight features)CNN model.Compared with the traditional algorithm,the enhanced feature CNN model not only has the advantages of the CNN model in image feature extraction,but also makes full use of the orders and business information in the restaurant food receiving process,which greatly improves the classification effect of the food image.In the context of receipt of restaurant ingredients,because the order and other business information are closely related to the type of ingredients,this article first designed the ResNet-Order model based on order characteristics.This model is the basis for enhancement features CNN model,and because each ingredient has its own unique weight distribution,this paper finally considers the characteristics of orders and weights comprehensively and designs a enhanced feature CNN model ResNet-Enhanced.The input of general CNN model is mainly image,order feature and weight feature cannot be used as the input of CNN model.To solve this problem,this paper designs a fully connected layer that can input order features and weight features,this provides great convenience for training enhanced feature CNN models.In addition,because the order features and weight characteristics in the restaurant food receiving process are unstructured data,this article also deals with these features,and the processed order features and weight features all have the same dimensions,providing training enhancement feature CNN model Data support.This paper proposes an image classification algorithm based on enhanced feature CNN model to make full use of the image features of food images and the order and weight features in the business process to make up for the lack of general CNN model,which consider only the features of food images,thus improving the classification accuracy of food images.The final experimental results show that compared with the general CNN model,the enhanced feature CNN model has further improved its classification accuracy and convergence speed.