Dog breed classification using transfer learning

Abstract

This research focuses on the use of machine learning and computer vision techniques to predict the dog breed using a set of images. In this, the convolutional neural network is used as a base which is responsible for the prediction of dog breed. For making this model effective, there was further inclusion of two cases. The first case was that to give the human image as an input, and it will provide the resemblance breed of the dog as output, and the second was to give an image of things other than a dog or a human, and it will provide “something else” as an output. The test accuracy of the model is 84.578%.