An Enhanced Convolution Neural Network Approach with Higher Classification rate for Images
DOI:
https://doi.org/10.17762/msea.v72i1.2397Abstract
Image categorization necessitates the development of features that can recognize image patterns that reveal group identity. The goal of this research is to classify photos from the public domain. Deep learning algorithms wereused to combine diverse picture feature sources from the CIFAR-10 image dataset. The majority of typical convolutional neural networks (CNN) have the same structure: convolutional layer modification and a max-pooling layer procedure coupled by a number of completely linked layers. The primary goal of this paper is to increase the efficiency of simple convolution neural network models. On a dataset from the Canadian Institute for Advanced Research (CIFAR-10), the Artificial Neural Network (ANN) technique isused with two distinct CNN topologies. After 10 hours of running, the updated model achieves an 88 percent classification accuracy rate. The Keras library, which is available for Python programming, is used to implement the deep learning