Skin cancer is a dangerous disease for human beings and it can spread immediately to other body areas. Therefore, it is of the utmost importance to make a skin cancer diagnosis early stage in order to save the patient’s life. The proposed strategy divides skin malignancies into two categories: melanoma and benign. In this study, the classification of skin cancer using four Deep Neural Network algorithms is proposed such as CNN, ResNet, RNN and DenseNet, respectively. The proposed deep learning algorithms with proper functions and tools are used for the classification and their comparative study on skin cancer. The dataset used is ISIC 2018. The main purpose of the work is to get the maximum accuracy and computational costs to detect an image among the four different Deep Neural Networks. By training on a significant dataset of labelled images of skin lesions, the proposed approach ensures generalisability across a range of skin malignancies. CNN, a subset of the Deep Neural Network, can learn the features of the input by using training and layers. ResNet is made up of residual blocks that have been through several conventional layers. DenseNet consists of a dense block of data that allows the construction of a very Deep Neural Network with fewer parameters and better gradient flow. RNN is a developed model to work with a sequential data set such as natural language processing and time series. The comparison of accuracy is made on all four algorithms in between CNN has achieved a maximum accuracy of 96%.