Transfer Learning in Convolutional Neural Networks for Fine-Grained Image Classification

Nicholas C. Becherer


In recent years, convolutional neural networks have achieved state of the art performance in a number of computer vision problems such as image classification. Prior research has shown that a transfer learning technique known as parameter fine-tuning wherein a network is pre-trained on different datasets can boost the performance of these networks. However, the topic of identifying the best source dataset and learning strategy for a given target domain is largely unexplored. Thus, this research presents and evaluates various transfer learning methods for fine-grained image classification as well as the effect on ensemble networks. The main contributions are a framework to evaluate the effectiveness of transfer learning, an optimal strategy for parameter fine-tuning, and a thorough demonstration of its effectiveness. The experimental framework and findings will help to train models in reduced time and with improved accuracy for target recognition and automated aerial refueling.