Skin is the largest and fastest-growing organ in the human body, which protects the body from the effects of the external environment. The skin is frequently exposed to the outside environment, so the possibility of the skin lesion is high. In recent years, the incidence of skin diseases is increasing rapidly. Some skin lesions develop into malignant tumours, sometimes we can see or feel, but others only detect through diagnostic imaging tests. The medical examination of skin lesions is not a simple task, as there is a similarity between lesions and medicals experience that can result in inaccurate diagnoses. Many cases of cancer are misdiagnosed as another disease with injurious consequences. In this article, we propose a method for classifying skin lesions using a shallow convolutional neural network (CNN). We perform the prediction tasks on a public International Skin Imaging Collaboration (ISIC) 2019 dataset and achieve a ROC-AUC of 0.782 for an eight-class classification of eight various types of skin.