Design Of Mushroom Recognition APP Based On Deep Learning Under Android Platform

The safety of food is always,Strengthening food safety concerns the health and safety of more than 1.3 billion Chinese people.All departments at various levels have paid great attention to food safety to ensure the safety of the people's "tongue".At the same time,article 11 of the food safety law of the People's Republic of China(2018 revision)explicitly encourages and supports the adoption of advanced technologies and management standards to improve food safety.According to incomplete statistics of the Chinese center for disease control and prevention in the past 10 years,the number of cases of accidental consumption of wild mushrooms in China accounts for 12% of the total number of cases of food poisoning in China,and the death toll is as high as 35%.That is to say,1/3 of the deaths of food poisoning in China are due to accidental consumption of wild mushrooms.Wild mushrooms have become the "number one killer" of food poisoning in China.In view of this situation,this paper proposes to apply the deep learning technology to the classification and identification of mushrooms,and realize a rapid identification of mushrooms on the mobile phone with the mobile platform.In this paper,a 13-layer convolutional neural network including input layer is built on Keras platform for end-to-end model training,which directly acts on the original image data of mushrooms.Through feature learning layer by layer,the feature information of mushrooms is obtained by using multi-layer network,thus avoiding the difficulties and problems of manual feature extraction.On this basis,the efficient and intelligent identification of mushrooms is realized by using the optimized objective function.In view of the similarity between mushrooms,especially those of similar genera,that is,to differentiate different kinds of mushrooms in fine grain size.The implementation of the system is divided into two stages: model training and offline recognition.Model training is completed on Windows platform and offline recognition is implemented on Android side.Because of the lack of necessary data sets,this paper firstly constructs eight different types of data sets by using network crawling technology.In order to train the network model,the sorted data sets are labeled manually.Then,under Keras learning framework,CNN convolution neural network is used to train and optimize the network model of image data set,and the network model is combined with APP in Android end to realize mushroom recognition in mobile end.Experimental debugging results show that the implementation of mushrooms can identify APP to eight different kinds of mushrooms for effective identification,the average recognition rate can reach 95%,the system will be able to make the recognition result of mushrooms,not only can give corresponding mushroom related knowledge introduction,implements the mushrooms on the mobile end intelligent recognition and popularity,the study of mushroom have certain value to application and research.