Early Detection Of Melanoma Skin Cancer Using Efficient Netb6

Among the various known skin cancers, Melanoma is the death-dealing skin cancer. Melanoma takes a significant share in the deaths reported by skin cancer. It forms in melanocytes, the skin cells that produce melanin, a pigment that gives the skin its color. It is essential to get the cancerous changes detected and treated before cancer has spread. When a person develops symptoms with Melanoma and consulted a dermatologist, the Dermatologist uses a dermoscopy technique to see patches on the skin more conspicuously. Dermatologist must put in a lot of effort, and the process can be time-consuming. Early detection of Melanoma includes the dermatologist level screening followed by a biopsy if the spread appears severe. The dermoscopy process is usually time taking and can be inaccurate as well. We proposed a profound learning way of solving the problem using the EfficientNetB6 Algorithm. The main motive to use EfficientNEtB6 is that when comparing Efficient Nets, including other subsisting CNNs at ImageNet, the Efficient Net models accomplished both higher precision and greater productivity over enduring CNNs. So, in view of this paper, we propose the Deep Learning method of using non-invasive computer vision-aided medical clinical diagnosis. The proposed system uses the Efficient- NetB6 algorithm trained on five-folds over a dataset of thirty thousand skin lesion images aiding to detect melanoma at an early stage, given the skin lesion image predicting the probability of it being Melanoma.