Skin cancer is well known and regarded as one of the most common types of cancer, killing millions of people worldwide. Detecting and categorizing cancer at an early stage can be advantageous, leading to a faster and higher success rate of therapy. Intelligent technologies are currently being used to classify skin lesions. The fundamental goal of our experimental research is to investigate biomedical skin cancer datasets in order to develop an effective approach for determining whether a cancer is malignant or benign. To train and categorize the dataset images, CNN (sequential), ResNet-50, Inception v3, and Xception models are employed. Two large and balanced datasets are collected for this purpose. One is used to compare the performance of employed model algorithms. Next, the selected model is again retrained on the second dataset for validation and generalization purposes. It turns out that the performance of Xception model is generalized and outperforms other models in accuracy. Experimental results are tabulated and graphed using accuracy and confusion matrix.