Plant Leaf Recognition System Based On Android Platform

With the development of neural network in recent years,it plays a more and more important role in the field of image recognition,speech processing and so on.In the direction of plant leaf recognition,traditional identification methods are inefficient and difficult to get into people’s vision.The wide popularity of the Android platform makes the recognition of plant leaves easily get into people’s learning process and daily life.The leaf dataset is consisted of some common plant in campus and botanical garden.By rotating the leaf into different angle to shoot,this can increase the number of the dataset.Also this can make the model is robust against rotation.The paper consists of three parts:First,the system design is designed into two parts: Android platform and the cloud server platform.The system architecture used the idea of modular organization.The design of the overall workflow of the system according to the characteristics of each module and its task.the allocation of the Android platform and the cloud server is rational which can improve the system`s efficiency.Second,image acquisition and preprocessing based on Android system.The complex background has great interference on the training and recognition for plant leaf recognition which can make the recognition accuracy rate very low.In order to increase the accuracy rate of the image with complex background,the system can preprocessing the images to extract the foreground based on the leaf color or by using a few part of manual interaction to mark the background of the image,then the processed image with only leaf itself can be submitted to the server.The server can make identification with the extracted images by convolution neural networks.With this method the can improve the system`s accuracy rate.Third,the server witch running the convolution neural network to identify the leaf and feedback to the user.Convolution neural networks can learn the feature of the image independently and make a extraction.The accuracy rate of the recognition is higher than the traditional feature extraction methods.My convolution neural network has 7 layers,and the trained model is deployed on the cloud platform,which contains multiple server nodes.To improve the speed of system recognition by using the advantages of cloud computing.In this paper,the plant leaf recognition system based on Android platform is implemented.and the paper give the test of the whole process of the system,and the actual effect of each module is demonstrated.