Monkeypox is a viral disease found in Central and West Africa, and it presents notable public health challenges due to its clinical similarity to smallpox. Recent outbreaks in over 40 countries have underscored the urgent need for quick and accurate diagnosis, especially in settings with limited resources where traditional diagnostic methods may be inadequate. This study explores the application of deep learning techniques to classify four specific types of monkeypox skin lesions. Transfer learning was employed to evaluate the performance of different models using a dataset of 770 skin lesion images. Various methods, including data augmentation and image preprocessing, were utilized to improve the effectiveness of the models. The results indicate that the custom SqueezeNet model was the most successful, achieving an accuracy of ${9 8. 8 5 \%}$ and an F1 score of ${9 8. 8 9 \%}$. These results surpass previous research benchmarks across four key evaluation measures. Additionally, the model was integrated into a web-based diagnostic tool capable of reliably detecting different forms of monkeypox. This study represents a significant advancement in utilizing artificial intelligence for disease detection, offering a practical approach for early and accurate monkeypox diagnosis. The proposed system has thex potential for seamless integration into clinical practice, aiding healthcare professionals in timely interventions.