Skin cancer is recognized as the most prevalent and extensively documented form of malignancy globally. It arises from the abnormal proliferation of melanocytic cells, commonly known as melanoma, triggered by exposure to ultraviolet radiation and genetic predisposition. Detecting a condition early greatly increases the chances of treatment being successful. However, the conventional biopsy method utilized for skin cancer detection is both invasive and painful, involving laborious laboratory procedures that consume substantial time. To mitigate these challenges, computer-aided diagnosis systems offer promise. In our work, we have developed four distinct models based on EfficientNetB1, EfficientNetB3, and EfficientNetB5. To conduct a comparative analysis, we assessed various cutting-edge techniques. The method we have proposed, employing EfficientNetB3 outperforms the majority of these techniques, achieving an overall test accuracy of 94%.