Abstract:
The viral infection known as Monkey-pox has recently shown a significant increase in its spread.
Diagnosing the disease poses a challenge for experts due to its resemblance to other illnesses,
particularly those related to smallpox. This study contributes to the advancement of Monkey-pox
disease detection through the introduction of a highly effective deep-learning methodology. We
present an innovative method employing deep learning algorithms to accurately and efficiently
identify pox diseases using image processing techniques. We evaluate several deep learning CNN
architectures (ResNet50, InceptionV3, DenseNet121, and EfficientNetB2) based on their accuracy
and computational efficiency. Employing data augmentation techniques enhances model
generalization and diminishes overfitting, enabling robust feature learning. Our findings
emphasize DenseNet121's superiority, achieving an outstanding 93.18% accuracy. DenseNet121
consistently demonstrates superior confidence rates and accuracy across disease categories,
showcasing its reliability in pox disease identification. Moreover, DenseNet121 exhibits faster
prediction times compared to other models, enhancing its suitability for practical implementation.
