Evaluating VGG16, ResNet50, and Xception for COVID-19 Detection via a Web-Based Deep Learning Interface

The global outbreak of COVID-19 has emphasized the need for fast and accurate diagnostic methods, especially in areas where advanced testing facilities are scarce or unavailable. This research proposes a deep learning-based method for the binary classification of chest X-ray images aimed at detecting COVID-19 infections. We evaluate and compare the performance of three prominent Convolutional Neural Network (CNN) architectures ResNet50, VGG16, and Xception trained on publicly available chest radiograph datasets. Additionally, a confusion matrix, training versus validation accuracy plots, and the ROC curve are employed to gain deeper insights into the model's effectiveness and diagnostic capability. To enhance usability, a user-friendly web interface has also been developed, enabling users to upload images and obtain real-time classification results. Experimental findings indicate that all three models perform well in distinguishing COVID-19 cases, with the Xception model achieving the highest accuracy.