Publication

Abstract : Accurate skin tumor identification and categorization are crucial for quick diagnosis and treatment in dermatological care. In our research, we developed a reliable model that successfully categorizes a wide range of skin tumors, obtaining astounding accuracy rates of 91% with DenseNet and 97% with EfficientNet. This feat highlights the advanced architectures’ capacity for synergy. Our dataset includes an extensive collection of skin tumor images that includes ten different categories, providing a comprehensive view of all types of skin tumors. These exceptional accuracy rates were attained via meticulous model training and validation, demonstrating the effectiveness of our methods. Our study makes a substantial contribution to dermatological diagnostics using computers. Both DenseNet and EfficientNet recognize complex characteristics in skin tumor images, allowing accurate classification. High accuracy is acheived by both models, which also make the prediction more versatile. Our study’s consequences go beyond the confines of academia. Dermatologists and other medical practitioners may use our method as a powerful diagnostic tool for correctly describing skin growths, eventually improving patient treatment and results.