Classification of animals species using convolutional neural networks: A comparative study with support vector machines

Abstract

This study aims to compare the performance of convolutional neural networks (CNNs) and support vector machines (SVMs) in the task of classifying animal species from images, focusing on three main categories: wolves, foxes, and wild dogs. Both models were built and trained on a dataset containing 3,000 images distributed evenly among the three categories. The results showed that the convolutional neural network (CNN) model (CNNs) achieved significantly superior performance with an accuracy of 97% compared to the support vector machine (SVM) that relied on HOG features with an accuracy of 82% and the support vector machine (SVM) that used features extracted from CNN with an accuracy of 94%. These results confirm that convolutional neural networks (CNNs) are the best choice for classifying complex images such as animal images, thanks to their superior ability to automatically learn hierarchical features. However, the study also showed that support vector machines (SVMs) can achieve competitive performance when provided with rich features extracted by CNNs, suggesting the possibility of using a hybrid approach in some applications.