Abstract COVID-19, a highly infectious disease caused by a severe acute respiratory syndrome, poses a significant threat as it can lead to fatalities within a matter of days. The current pandemic necessitates extensive testing, which is a laborious and time-consuming process. Recent advancements done in Deep Learning, particularly in the field of image analysis, have proven to be effective. This study proposes and investigates the performance of three Convolution Neural Networks (CNNs) utilizing transfer learning and compares them against other existing architectures. To conduct the experiments, a publicly available dataset consisting of 3,792 Chest X-Rays categorized into three categories was employed: COVID’19 patients (labeled as Covid), patients with a negative diagnosis (labeled as Normal), and those with pneumonia. The chosen architectures for evaluation were vgg16, resnet50, and a custom CNN. Additionally, ensemble models were constructed and tested using various combinations. The findings demonstrated that the ensemble models consistently yielded the most favorable outcomes. Furthermore, all three CNN architectures exhibited remarkable performance, achieving an average accuracy of 97.7%.
Alan : Mühendislik
Dergi Türü : Uluslararası
Benzer Makaleler | Yazar | # |
---|
Makale | Yazar | # |
---|