Skin cancer ranks as one of the most common types of cancer worldwide, and detecting it early is essential for successful treatment. Deep learning techniques, particularly vision transformers, have demonstrated significant potential in enhancing the precision of detecting skin cancer. For our study, we employed an altered variation of the VGG16 and the Vision Transformer (ViT_B_16) pre-trained models and fine-tuned them using the ISIC dataset available on Kaggle. These models were chosen to explore their efficacy in classifying skin cancer images through advanced image processing techniques. The findings from the experiments show that the altered VGG16 model attained an accuracy of 86.82%, whereas the ViT_B_16 model exceeded this with an accuracy of 90.75%. Both models maintained a similar size, demonstrating the efficiency of the vision transformer architecture in handling complex image-based tasks. The top-of-the-line performance of the ViT_B_16 model highlights its capability as a strong instrument for identifying skin cancer. Future work will focus on further optimizing these models and exploring their application in practical clinical environment to enhance diagnostic procedures.