Flower image classification based on improved convolutional neural network

This paper attempts to apply neural networks to classify flower images. The construction of the classification model is based on thousands of actual flower pictures, including Daisy, Dandelion, Rose, Sunflower, and Tulip. Convolutional Neural Network, or CNN, is often used in the analysis of visual images and has a good application in the classification and prediction of flower images. This paper analyzes the models and algorithms used in our experiments. By improving and optimizing the data set and neural network structure, we have achieved a better model structure and further enhanced the accuracy of flower prediction. The final training accuracy rate exceeded 91%.