Images Classification of Dogs and Cats using Fine-Tuned VGG Models

Autor: Mahardi, I-Hung Wang, Kuang-Chyi Lee, Shinn-Liang Chang
Rok vydání: 2020
Předmět:
Zdroj: 2020 IEEE Eurasia Conference on IOT, Communication and Engineering (ECICE).
DOI: 10.1109/ecice50847.2020.9301918
Popis: Image classification has become more popular as it is the most basic application and implementation of deep learning. Images of dogs and cats are the most common example to train image classifiers as they are relatable. It is easy to classify the image of cats and dogs, but the images of various breeds are difficult to classify with high accuracy. In this paper, we tried to build an image classifier to recognize various breeds of dogs and cats (CDC) using fine-tuned VGG models. Two common models, VGG16 and VGG19 were used to build the classifier. The resulting model from VGG16 has a training accuracy of 98.47%, validation accuracy of 98.56%, and testing accuracy of 83.68%. The model from VGG19 has a training accuracy of 98.59%, validation accuracy of 98.56%, and testing accuracy of 84.07%.
Databáze: OpenAIRE