Performance and Accuracy Analysis of the ResNet50 Transfer Learning Model for Melanoma Cancer Classification

An important component in the treatment of severe skin diseases is the early diagnosis of melanoma cancer. To assess the effectiveness and accuracy of the model, this paper analyzes the classification of melanoma cancer using the ResNet50 model. ResNet-50 increases the possibility of detecting melanoma cancer at an early stage with the combination of its strong feature extraction capabilities and deep structure. It also increases the chances of successful treatment and a better prognosis. In the dataset of photos labeled “benign“ and “malignant”, augmentation techniques such as rotation, zoom, vertical inversion, and horizontal inversion are used to improve model performance. The results showed that the model with a calculation time of one hour and forty-five minutes achieved $86 \%$ accuracy. By utilizing this augmentation strategy, the model’s capacity to identify melanoma cancer cases has been enhanced, thereby making a significant contribution to the field of cancer detection research. This study offers a new perspective on the use of the ResNet50 model for melanoma cancer classification and raises the possibility of using related approaches to address other health problems.