Intelligent Animal Classification System Using Convolutional Neural Network

Abstract

The usage of intelligent software to distinguish and arrange protests in computer vision is an innovation of developing significance in numerous fields, including wildlife conservation and management. Deep convolution networks are achieving a tremendous performance in object detection and classification. The current study proposed a deep architecture of convolution neural network for the purpose of improving the accuracy of identifying the various animals. Still now there is no unique method to classify the animal images efficiently using CNN. Here we classified the animal images using different popular deep networks based on CNN like LeNet, AlexNet,VGGNet and ResNet. These kinds of networks worked efficiently for object classification task but needed sliding window concept for localizing object in the image. It works gradually slow as it needed to process many windows for the single image. So we have combined the YOLO algorithm to improve the object detection in real-time with high speed. This YOLO algorithm applies a single neural network to the complete full image, and then divides the image into large number of blocks and estimates the bounding boxes along with their probabilities for each block accurately. We provided a detailed explanation of how these algorithms work and comparison between them.