Abstarct: Abstract This thesis utilizes modern deep learning methods for the classification and segmentation of skin cancer. Skin cancer is one of the most common types of cancer, and early and accurate detection can play a crucial role in reducing mortality rates associated with this disease. Due to the complex nature of medical skin images, advanced techniques are required for more precise identification and analysis of these images. For this research, the ISIC2020 dataset was used, which is one of the most reputable datasets in the field of skin medical images. One of the main challenges in this study was the imbalance in the dataset, which led to some types of cancer being underrepresented compared to others. Additionally, the available datasets lacked accurate masks for image segmentation. To address these challenges, segmentation masks were first generated using deep learning methods, and then, data augmentation techniques were applied to balance the dataset. For the segmentation task, the DoubleU-Net model was employed. This model is specifically designed for precise medical image segmentation and demonstrated excellent performance. The model was evaluated using the Dice coefficient, achieving an accuracy of 98.09% in identifying cancerous skin regions, indicating its high efficiency in this task. For the classification task, the ResNet-50 model was used, enhanced with pre-trained weights, multi-scale techniques, and attention mechanisms. This model was able to identify important features of the images with high accuracy and demonstrated excellent performance in classifying skin cancer types. The experimental results showed that the model achieved an accuracy of 91.01% on test data and 96% on training data.