This study introduces a practical and innovative approach to Monkeypox diagnosis using advanced deep learning (DL) and transfer learning (TL) techniques. Eight pre-trained models were fine-tuned and assessed for their performance in identifying Monkeypox from medical images. VGG19 and MobiletNetV2 emerged as the best-performing models with accuracies of 99%, respectively. To enhance trust and usability, the study incorporates Local Interpretable Model-Agnostic Explanations (LIME), making the model predictions transparent by highlighting critical image features influencing the diagnosis. This transparency is crucial to helping healthcare providers in decision-making. Futher, lightweight models like VGG19 are optimized for use in resource constrained settings, improving accessibility in undeserved regions. By addressing challenges such as overfitting through data augmentation and testing on datasets of varying sizes, the study demonstrates the potential of DL and TL to deliver accurate, interpretable, and scalable solutions for early Monkeypox detection.