Image Classification with Deep Learning and Comparison between Different Convolutional Neural Network Structures using Tensorflow and Keras

Abstract

Deep learning technologies are becoming the major approaches for natural signal and information
processing, like image classification, speech recognition. Deep learning is a technology inspired by the functioning of
human brain. In deep learning, networks of artificial neurons analyze large dataset to automatically discover underlying
patterns, without human intervention, deep learning identify patterns in unstructured data such as, Images, sound, video
and text. Convolutional neural networks (CNN) become very popular for image classification in deep learning; CNN’s
perform better than human subjects on many of the image classification datasets.
In this paper, a deep learning convolutional network based on keras and tensorflow is deployed using python for binary
image classification. In this study, a large number of different images, which contains two types of animals, namely cat
and dog are used for image classification. Four different structures of CNN are compared on CPU system, with four
different combinations of classifiers and activation functions.
It is shown that, for Binary image classification combination of sigmoid classifier and Relu activation function gives
higher classification accuracy than any other combination of classifier and activation function.