A Comparative Performance Analysis of Various Pre-Trained Convolutional Neural Network Models

Abstract

The forms of the data are changing with technological advancement. With this revolution, the text data sets were replaced by images and today technological advancement has resulted in large datasets in the form of videos. However, there arises the need of technology to deal with images and develop intelligent systems that can identify required information correctly from the images. Such task can be accomplished by the pre-trained models of convolutional neural network. In this paper, we come up with a comparison of performance of various pre-trained cnn models like alexnet, googlenet, squeezenet etc. that are used for image classification. Here, we compare and display the accuracy level of these cnn models of detecting an object from a huge dataset. The performance has been measured on a single dataset of flowers available in the online repository of Kaggle.