Flower Recognition Based on Convolutional Neural Network

In view of the problem that flowers still need to rely on experience to manually select features and low recognition rate in complex background at present, an 8-layer convolutional neural network is designed and compare the influence of various parameter selection on network performance during network construction, and a recognition method of deep convolutional neural network is proposed. Through the convolutional neural network learning feature, the features of the last layer in the network are integrated and connected as the output feature vector, then the characterization data was trained and tested using a Random Forest classifier. The Oxford 17 Flowers dataset is augmented, then classified, and compared with other classifiers. The experiment showed that the recognition rate of Random Forest classifier reached 81.37%, which is significantly better than Softmax classifier and Support Vector Machine.