Convolution Neural Network based Transfer Learning for Classification of Flowers

Abstract

Flower plays an extremely important role in our life, which has high research value and application value. The traditional methods of flower classification is mainly based on shape, color or texture features, and this methods needs people to select features for flower classification lead to the accuracy of classification is not very high. This paper aims to develop an effective flower classification approach using convolution neural network and transfer learning. In this paper, based on VGG-16, VGG-19, Inception-v3 and ResNet50 models were used to compare the network initialization model with the transfer learning model. The results show that transfer learning can effectively avoid deep convolution networks are prone to local optimal problems and over-fitting problems. Compared with the traditional methods, the accuracy of flower recognition on Oxford flowers dataset is obviously improved, and has better robustness and generalization ability.