Skin cancer is a hazardous ailment and a leading contributor to mortality. Early diagnosis of skin cancer can significantly minimize or prevent these fatalities. The diagnostic process can be both lengthy and costly. This study aims to establish a basis for creating an exact and effective convolutional neural network (CNN) model for detecting skin cancer. This model aims to enhance the early and accurate diagnosis of skin cancers, potentially leading to a decrease in the number of associated fatalities. The model utilizes a convolutional neural network with Keras Tensorflow as the backend to categorize seven distinct categories of skin cancer. Subsequently, the results are analyzed to determine the practical applications of the model. The MobileNet optimizer is used for classification. A web application has been developed that offers dermatologists the three most likely diagnostics for a specific blister. It will aid in swiftly recognizing patients with high priorities and accelerating their workflow. The application generates a result within a time frame of 5–10 seconds.