A Convolutional Neural Network Approach to Picture Classification

Abstract

Due to the rapid advancements in digital content identification in the last year, automated picture categorisation has emerged as the most difficult problem in computer vision. When compared to human vision, automatic image comprehension and analysis by systems is challenging. Despite several attempts to address the shortcomings of the current categorisation method, the results have been limited to crude images at the lowest level. Nevertheless, the accuracy of picture categorisation is lacking in such methods. Our system achieves predicted outcomes in areas like computer visions using a deep learning approach, as described in this work. In order to automatically categorise the photos, our system use a machine learning method called Convolutional Neural Network (CNN). In order to classify greyscale photos, our system compares itself to the Digit of MNIST dataset. More processing capacity is needed for picture categorisation on the basis of the greyscale photos included in the training data set. Our model achieves great accuracy in picture classification, as shown by the 98% accuracy result in the experimental portion, which was obtained by training the images using CNN networks.