Image classification using Convolutional Neural Network
No Thumbnail Available
Date
2022
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Image classification is a challenging task in computer vision, great progress has been achieved
in recent years due to the application of methods of deep learning, especially the Convolutional
neural network (CNN). In this work we suggest four convolutional neural network models
trained using the intel image classification challenge dataset, this dataset consists of natural
scene images from over the world it contains around 25 000 images distributed under six categories,
there are around 14 000 images in the train set, 3 000 in the test set and 7 000 in prediction
set. Our four models differ in the number of hidden layers (10 to 18), we took this approach
seeking to achieve the best results, and we have applied some techniques to improve the performance
of our convolutional neural network models as data augmentation, dropout, and batch
normalization. Our solution produces good results in terms of accuracy and performance.