Medical diagnosis is crucial for saving lives, and deep learning models enhance accuracy, particularly in early disease detection like pneumonia. Swift treatment is vital as pneumonia can be life-threatening, especially for vulnerable groups. Despite significant advancements in medical imaging, accurately detecting pneumonia in X-ray images remains challenging due to their complex nature.

To address this challenge, we utilized a Chest X-ray dataset to apply deep learning models for pneumonia detection, assessing four well-known networks: AlexNet, ResNet18, GoogleNet, and VGG16. Our primary aim was to enhance diagnostic accuracy and efficiency. Results indicate that transfer learning notably boosts performance, with ResNet18 achieving the highest test accuracy of 87%. Additionally, we explored the applicability of Generative Adversarial Networks (GANs) for X-ray image classification, surpassing traditional methods. These findings demonstrate the potential of deep learning models and GANs to rapidly and accurately identify pneumonia in X-ray images, promising better health outcomes.