A Study of Gaussian Fuzzy Preprocessing for Image Classification

The evolution of artificial intelligence has progressed through distinct waves since the 1960s, with a current focus on deep learning. It is worth noting that, due to the development of computer vision, machines are now able to process visual data in a similar way to humans. In healthcare, artificial intelligence integration, particularly in diagnostics, leverages machine learning and deep learning to enhance speed, accuracy, and early disease detection, significantly elevating healthcare quality. This study aims to develop a system integrating fuzzy preprocessing with GoogLeNet image classification. Initially, publicly available image datasets, the COVID-19 Lung CT Scans Dataset are utilized. These datasets are divided into training, validation, and test sets. Gaussian fuzzy preprocessing is applied to the images, followed by training and validation of the GoogLeNet model. Subsequently, the trained model is tested on the dataset, achieving an average accuracy of 89.87%. Furthermore, if this technology is applied to future manufacturing processes, it can lead to higher production efficiency and improved product quality.