Abstract: The purpose of this research is to present a unique artificial intelligence model for image processing by leveraging data cleaning and optimization techniques in a deep convolutional neural network. This novel trifecta of methods has the potential to improve the performance of deep neural networks in image processing. Priority was given to the often-overlooked and inadequate data cleaning and optimization stages required when constructing a deep neural network. We were able to get positive outcomes by employing this strategy. We put this conceptual framework to the test by comparing the chest X-rays of people with and without pneumonia to look for commonalities. Several performance metrics, including precision, recall, and F1-score, were enhanced, and the true positives and true negatives were both improved. The accuracy improved, and we saw a notable enhancement while employing our proposed framework. All things considered, the outcomes imply that our framework provides superior results without compromising any performance indicators.
Keywords: image classification, convolutional neural network, deep learning, VGG16, Inception v3, Xception
Copyright Form