LAUSR

Machines can acknowledge the images of skin lesion as well as the disease compared to an experienced dermatologist. These might be executed by giving a proper label for the provided images of skin lesion. Within the proposed study researchers have examined various frameworks for detection of skin cancer as well as classification of melanoma. The current research includes a unique image pre-processing technique and modification of the image followed by image segmentation. The 23 texture and ten shape features of the dataset are further refined with feature engineering techniques. The improved dataset has been processed inside a Deep Neural Network models by binary cross-entropy. The dataset passes through several mixes of multiple activation layers with varying features and optimization techniques. As an outcome of the study, researchers have selected a useful, timesaving model to find an image as melanoma or even naevus. The model was evaluated with 170 images of MED NODE and 2000 images of ISIC dataset. This improved framework achieves a favorable accuracy of 96.8% with few noticeable epochs which concern other 12 machine learning models and five deep learning models. In the future, certainly there can be an investigation with several classes of skin cancer with an improved dataset.