Research And Implementation Of Poisonous Flower Identification And Classification Based On Hybrid Mode

Computer vision includes many aspects,and image classification is one of the hot topics.Image classification can be understood as the process of using computers and using algorithms to find the category labels of images and distinguish the categories of different images.Image classification has the following steps: image preprocessing,feature extraction and classifier classification.In these steps,feature extraction is particularly important,which determines the accuracy of image classification.In the traditional image classification task,the texture and color of the image are often extracted for judgment.This method has a relatively ideal effect on the classification of simple images,but it cannot be achieved for the classification of complex images.Therefore,image classification algorithm based on convolution neural network with better classification effect for complex images has gradually become a focus of attention.Based on it,the paper proposes to use the advantages of convolutional neural network(CNN)to extract image features,and combine with support vector machine(SVM)to design a new hybrid model to reduce the time of image training and improve the accuracy of image recognition.Applying this new hybrid model to the research of image classification and recognition can complete the classification and recognition of poisonous flowers,which verifies the advantages of the hybrid model.The main research contents of this paper are as follows:(1)First of all,selecting a variety of types of flowers to be classified into two categories: toxic and non-toxic.Collecting the images of these flowers through the methods of downloading the data set published on the network and taking photos on the spot,then enhancing and pre-processing the images to establish flower image data set.(2)Then compared the advantages of different convolutional neural networks,such as Le Net,Alex Net,VGG,and so on,selected the Alex Net network model to improve it to extract image features,optimize the activation function,fuse support vector machines to form a hybrid model,and apply the hybrid model to the classification and recognition of toxic flowers.The mixed model was trained and tested,and a comparative experiment was conducted with the original model.The experimental results showed that the mixed model had faster convergence speed and higher recognition accuracy.(3)Finally,the network model and the training and testing system are compiled with the pytorch framework,and the web page is compiled with the django framework to achieve visual interaction.Through the designed hybrid model,the functions of inputting flower pictures to get the name of the flower and whether the flower is poisonous are realized.The experimental results show that the classification effect of the new hybrid model for toxic flowers is better than the original model,and the new hybrid model improves the recognition accuracy,which is more robust.The classification and recognition system for toxic flowers based on the hybrid model has certain practical value and research significance in preventing the occurrence of flower poisoning events,maintaining ecological diversity,and improving the work efficiency of scientific researchers.