Dataset Construction And Classification For Chinese Cuisine Ingredients

Image classification is by using the method of image feature to distinguish the different image category,the traditional image classification method by using the method of feature descriptor add classifier,along with the development of artificial intelligence and deep learning technology,convolution neural network more and more application in the image classification,and showed a good classification performance.In the image classification task,the accuracy of image classification can be effectively improved by optimizing the network model and enhancing the data set.At present,most models are optimized only for specific data sets,and it is usually necessary to transform the model when the established model is applied to incremental data sets.Although there are a large number of public data sets available for image classification at present,targeted data sets are very limited.Currently,the publicly available data sets that can be used for dish classification are very scarce.In this paper,a dataset of 369 categories of food materials is constructed for the above problems,and a multi-layer convolutional neural network model is proposed and analyzed accordingly.The specific research content includes the following points:First of all,a large-scale Chinese food ingredients dataset is constructed,including a total of 369 ingredients category.The web crawler technology is mainly applied in the construction process of the data set.The specific construction process is to determine the list of food materials in the data to be constructed by crawling technology and statistical techniques.According to the determined list,the three-level web crawler technology is used to obtain the images,and the images are screened by semi-automatic processing,and finally a large-scale food data set containing 369 categories is constructed.Secondly,In this paper,a multilayer convolutional neural network model is proposed.When constructing the convolutional neural network,the spatial transformation layer is adopted at the very beginning of the construction of the convolutional neural network,so that the convolutional neural network has an explicit image processing module,which can ensure the invariance of image data in the process of model learning.Finally,In this paper,a multi-group comparison experiment is designed to verify the effect of constructing the multi-layer convolutional neural network model in the image classification task.The experimental results show that the model constructed in this paper has a good effect on the image data set of Chinese dishes.Compared with the current mainstream convolutional neural network,,the model constructed in this paper has achieved good results in the classification and comparison of the food data set constructed in this paper.