Melanoma Moles Detection and Classification Using Cornernet Model

Melanoma is a malicious form of skin cancer and can develop wounds on the entire human body which can lead to the victim's demise at the advanced level. The timely and accurate detection of melanoma moles can prevent the victims from painful treatment and surgeries. However., immense similarity among the healthy and diseased areas and massive variations in the mass, contour., and shades of the melanoma moles complicate the recognition process. Here., we have employed a deep learning (DL) framework named the CornerNet model for the localization and categorization of skin lesions. The Hourglass-oriented CornerNet approach is used for deep features calculations which are later grouped by the one-step object recognizer network of the employed solution. The approach is tested on a standard data sample called the ISIC-2016 database and secured the average mAP and IOU values of 0.916., and 0.921 with an accuracy value of 90.08%. The model testing confirmed that our solution is vigorous for melanoma mole detection and cataloging.