An Efficient Approach towards Skin Cancer Diagnosis with EfficientNetB3

Skin cancer has been identified as the most widespread and well-documented type of malignancy worldwide. Its origin lies in the irregular growth of melanocytic cells, often referred to as melanoma. Exposure to ultraviolet radiation and genetic factors leads to melanoma appearing on the skin. Identification at an early stage increases the chances of successful treatment. However, the conventional biopsy method used for detecting skin cancer is both invasive and painful. It involves extensive laboratory procedures that consume a considerable amount of time. Computer-aided diagnosis systems can help to address these challenges. In this work, two distinct models has been developed based on EfficientNetB3, with varying additional layers. To conduct a comparative work, various cutting-edge techniques have been evaluated. The suggested approaches surpass the majority of these techniques by achieving an overall test accuracy of 91% and 93.25% for Model 1 and Model 2 respectively.