Abstract: Melanoma is the most lethal type of skin cancer in the world. Several attempts have been made to detect melanoma early using deep learning techniques based on dermoscopic images. For an accurate diagnosis of melanoma, it is important to distinguish complex lesion patterns. The typical lesion patterns, on the other hand, are not consistently present, resulting in sparse labelling issues in the results. We suggest a multi-tasking system in this paper

Keywords: opencv, feature extraction, image pre-processing, CNN algorithm etc.

| DOI: 10.17148/IARJSET.2021.85123