In medical imaging, accurate pneumonia detection from chest X-rays is vital for timely diagnosis and treatment. This study evaluates four deep learning models—Simple CNN, DenseNet121, VGG16, and InceptionV3—using the Kaggle “Chest X-Ray Images (Pneumonia)” dataset of 5,863 images categorized as normal and pneumonia. Data preprocessing involved normalization and augmentation, with model evaluation based on accuracy, precision, recall, and F1-score. Simple CNN achieved the highest accuracy (92%) with strong precision (0.95 normal, 0.90 pneumonia) and recall (0.83 normal, 0.97 pneumonia). VGG16 scored 91%, while DenseNet121 and InceptionV3 performed lower, the latter with 84% accuracy. Simple CNN was deployed in a Django-based web app on AWS for clinical use, supporting healthcare professionals at Dr. Sumait Hospital.
