A Study Based On Deep Convolution Neural Network For Flower Classification

With the development of big data in recent years,and the development of hardware such as GPU,containing more hidden deep convolutional neural network layer has more complex structure.Compared with the traditional machine learning method,they have more powerful features extraction and stronger expression ability.Now people can take photos of flowers easier through mobile phone,iPad,digital cameras and other mobile devices whenever and wherever.But most of them don't recognize the professional classification of various kinds of flowers,flowers recognition still needs to be carried out in professional knowledge and guidance.Because of the complexity of the environment,the similarity of flowers and the difference of the same kind of flowers,the traditional image classification methods can't solve these problems well.Therefore,this paper studies the images recognition of flowers by the multi-level convolution neural network.The main work and Innovation:Based on the traditional machine learning method for classification of flowers,it has a high dependence on the image segmentation effect and the manual feature selection,and the accuracy is not very high,this paper proposes to use convolutional neural network method for classification of flowers.The final experimental results show that the more deep convolutional neural network layer for feature extraction and classification has more advantages than the traditional methods.It does not need the flowers on the image segmentation and image feature selection can be processed by computer,and the final recognition rate than the traditional the method has been greatly improved.