Skin Cancer is the most common (accounting for 40% of cancer cases globally) and potentially life-threatening type of cancers. It was diagnosed in about 5.6 million individuals last year. Automated classification of skin lesions through images has been a challenge throughout the years because of fine variability in their appearance. Deep Learning techniques exhibit potential in tackling fine-margined image-based analysis and manage to provide accurate results. The three modelling stages include data collection and augmentation, model architecture and finally prediction into 7 different types of skin cancer namely actinic keratoses, basal cell carcinoma, benign keratosis-like lesions, dermatofibroma, melanoma, melanocytic nevi and vascular lesions. A Convolutional Neural Network was fabricated (using TensorFlow) obtaining an accuracy of 81.24%. Further Transfer learning Approach was implemented in PyTorch, which yielded accuracies of 96.40%, 98.20%, 98.70% and 99.04% respectively for Wide Resnet101, Resnet50, Densenet121 and VGG19 with batch normalization, which are all trained end-to-end from images directly, to proliferate the scalability of these models and curtail initial diagnostic costs. The aim of this research paper is to render non-invasive skin cancer screening a common norm, making it simpler. © 2020 IEEE.