Monkeypox has recently emerged as a significant public health challenge, necessitating efficient diagnostic methods for timely detection. This study addresses the critical need for rapid and accurate diagnosis of monkeypox by leveraging artificial intelligence and transfer learning techniques, aiming to overcome limitations in current diagnostic methods. A publicly available dataset from Kaggle, comprising 228 original images labeled as monkeypox and other skin conditions, was augmented to 3192 images to address data scarcity and enhance model robustness. Pre-trained deep learning models, VGG19 and EfficientNetB4, were employed to extract image features, which were then classified using the XGBoost algorithm, known for its effectiveness in structured data classification. The proposed approach achieved high accuracy rates of 100% and 97.02% for the original and augmented datasets, respectively. Additionally, 5-fold cross-validation results demonstrated accuracies of 85.98% for the original dataset and 93.42% for the augmented set, highlighting the model's strong generalization capabilities. The proposed approach combines transfer learning with ensemble classification, providing a scalable solution that shows improved performance over several existing diagnostic methods in terms of accuracy and computational efficiency. The findings underscore the transformative potential of AI-driven diagnostic tools in public health, paving the way for more rapid, accessible, and accurate detection strategies for monkeypox and other emerging infectious diseases.