Skin cancer that evolved in melanocytes, and early identification is crucial due to its ability to unfold. Proposed research affords a unique technique to cancer detection using deep gaining knowledge of methods and switch mastering methodology. Convolutional neural networks (CNNs) are proposed as a possible approach for distinguishing benign from malignant pores and skin lesions. It evaluates overall performance in pores and skin lesion categorization the usage of four alternative deep mastering architectures such as MobileNetV2, ResNet50, InceptionNetV3, and VGG19. Proposed study exceptional-tunes and trains fashions using a numerous variety of dermoscopic pictures a good way to obtain accuracy, computational performance, and processing speed for actual-world clinical use. In this research additionally highlights the influence of version structure on classification metrics which affords treasured insight into the satisfactory line among accuracy and use of computation aid in pores and skin lesion prognosis.