| Popis: |
Skin lesions are a common sign of many human diseases and are a worldwide indicator of many different types of cancer. The necessity of such skin cancer preventive initiatives is highlighted by, increased risks brought on by the effects of climate change and by the high expense of treatment. The early detection of Skin Cancer can be done to save many lives. Melanoma is the deadliest type of Cancer out of the known types so far. HAM 10000 and ISIC 2019 are the datasets that are used to classify seven and eight classes in the proposed article. The approach is five-layer CNN. The impact of data augmentation was analyzed using the proposed framework in two different datasets. Compared to the original data, the evaluation metrics for augmented data are high. For the ISIC 2019 and HAM10000 dataset, the CNN fine-tuned 5-layered model with augmentation achieved 98.67, 97.88 percent accuracy. |
