Skin cancer, such as melanoma, basal cell carcinoma, and squamous cell carcinoma, is a big world health problem because it is becoming more common and can be deadly if it is not found and treated. Finding cancer early is very important for a better patient’s outlook, but doctors have to examine patients by hand, which can take a long time and lead to mistakes. Recent progress in deep learning (DL) has made it possible to automate the study of dermoscopy pictures, which are commonly used to find skin cancer. Deep learning methods, especially convolutional neural networks (CNNs), have shown that they can improve the accuracy of diagnoses. This could lead to early discovery and less human mistake. This essay looks at different deep learning methods for using dermoscopy pictures to find skin cancer. First, we’ll talk about dermoscopy, a non-invasive imaging method that doctors use to get early diagnoses of skin problems. The main topic is how to use CNN-based models that are learnt on big sets of labelled dermoscopy pictures. These models learn complicated patterns and features from the pictures on their own, which lets them tell the difference between cancerous tumours and normal blemishes very accurately. The study looks into various deep learning designs, such as standard CNNs, deeper structures like ResNet and Inception networks, and new mixed models that mix CNNs with machine learning methods like transfer learning and reinforcement learning. Transfer learning is a way to improve model performance by fine-tuning models that have already been trained on skin cancer datasets. This works especially well when labelled data is scarce.