DESIGN AND DEVELOPMENT OF SKIN CANCER DETECTION USING VGG16 MODEL IN CONVOLUTIONAL NEURAL NETWORK

Abstract: Skin cancer is one of the most deadly types of skin cancer. If it is not recognised and treated right away, it is not anticipated to spread to other body parts. It also occurs when the tissue is exposed to sunlight, mostly as a result of the fast division of skin cells. A precise automated system for skin lesion recognition is crucial for early diagnosis since it can save time, effort, and human lives. Deep learning and image processing are both utilised in the effective treatment of skin cancer. The research proposes a computerised approach for categorising skin cancer. The seven main types of skin cancer have been classified in this study. Convolutional neural networks (CNN) with deep layers exhibit both efficacy and capacity. Seven clinical forms of skin cancer are represented in the dataset, including nevus, actinic keratosis, benign keratosis, dermatofibroma, and melanoma. The objective is to develop a model that can recognise skin cancer and classify it into several categories using a convolution neural network. The diagnostic approach makes use of deep learning and the idea of image processing. Through the use of different picture enhancing techniques, the amount of photographs has also grown. The accuracy of the classification tasks is then further improved by using the transfer learning technique. Accuracy rates are about 98 percent, according to the suggested CNN technique.

Keywords: Deep Learning, Convolution Neural Network, Melanoma or benign, Skin Cancer.