Classification of dog breeds using convolutional neural network models and support vector machine

Abstract

Convolutional neural network (CNN) has been widely used for fine-grained image classification, which has proven to be an effective approach for the classification and identification of specific species. For breed classification of dog, there are several proposed methods based on dog images, however, the highest accuracy rate for dogs (about 93%) is still below expectations compared to other animals or plants (more than 95% on birds and more than 97% on flowers). In this study, we used the Stanford Dog Dataset, combined image features from four CNN models, filtered the features using principal component analysis (PCA) and gray wolf optimization algorithm (GWO), and then classified the features with support vector machine (SVM). Eventually, the classification accuracy rate reached 95.24% for 120 breeds and 99.34% for 76 selected breeds, respectively, demonstrating a significant improvement over existing methods using the same Stanford Dog Dataset. It is expected that our proposed method will further serve as a fundamental framework for accurate classification of a wider range of species.