<![CDATA[Early detection of skin diseases remains a major challenge, particularly in regions with limited access to dermatological services. This issue is further exacerbated by the shortage of medical specialists and the widespread presence of inaccurate health information online. This study aims to develop an automated image-based classification system capable of identifying five types of skin diseases: Eczema, Melanocytic Nevus, Melanoma, Benign Keratosis, and Basal Cell Carcinoma. The proposed method utilizes a Convolutional Neural Network (CNN) with the VGG19 architecture, enhanced through transfer learning and partial fine-tuning at the block4_conv1 layer. A dataset of 10,000 JPG images was used, with preprocessing steps including normalization, data augmentation, edge detection, and class balancing. Model performance was evaluated using accuracy, precision, recall, F1-score, and confusion matrix. Experimental results show that the model achieved an accuracy of up to 84% in the best scenario, with balanced performance across other metrics, indicating strong multiclass classification capabilities. These findings demonstrate the effectiveness of VGG19 in detecting skin diseases from images. The results also suggest the potential development of mobile-based early detection systems to support communities in underserved areas.]]>
Copyrights © 2025
