IMAGE CLASSIFICATION USING CONVOLUTIONAL NEURAL NETWORKS

Abstract

In recent year, with the speedy development in the digital contents identification, 
automatic classification of the images became most challenging task in the fields of 
computer vision. Automatic understanding and analysing of images by system is 
difficult as compared to human visions. Several research have been done to overcome 
problem in existing classification system, but the output was narrowed only to low 
level image primitives. However, those approach lack with accurate classification of 
images. In this paper, our system uses deep learning algorithm to achieve the expected 
results in the area like computer visions. Our system present Convolutional Neural 
Network (CNN), a machine learning algorithm being used for automatic classification 
the images. Our system uses the Digit of MNIST data set as a bench mark for 
classification of grayscale images. The grayscale images in the data set used for 
training which require more computational power for classification of images. By 
training the images using CNN network we obtain the 98% accuracy result in the 
experimental part it shows that our model achieves the high accuracy in classification 
of images.