Evaluating Performance of Deep Learning Architectures for Image Classification

Abstract

VGGNet is a significantly more accurate CNN architecture that is more recently introduced. There has always been a question of which neural network architecture performs well in which scenario. Thus, using the Fashion MNIST dataset, the performance of VGGNet and CNN deep learning architectures is reviewed, and the metrics are compared. A 3 Layer CNN architecture was used in this work to achieve 98.92% training accuracy and 0.02 training loss and a maximum test accuracy of 90.77% in classifying 10000 images of 10 different types.