Deep Learning-based Classification for Melanoma Detection using XceptionNet and DenseNet

Melanoma is among the deadliest tumours, according to clinical trials and epidemiological research. It is now known that this condition is caused by the uncontrolled proliferation of melanocytes. General learning models and deep learning models are two more examples that have been used to reevaluate this paradigm. Experimental studies of melanoma also need visual data, such as dermatological scans or standard camera photos. It is more evidence that a trustworthy method for melanoma diagnosis is urgently needed. Some models that have been successful in melanoma diagnosis include Resnet, DenseNet, Inception, and VGG. Regardless, data is often only labelled as benign or malignant. Our study's data set consists of seven lesions: melanoma, dermatofibroma, basal cell carcinoma, actinic keratosis, vasculopathy, and blepharocytic nevus. We used a variety of deep learning models using the HAM10000 dataset. Confidence is bolstered by positive outcomes. Cancers of the skin account for at least 40% of all cancers and are among the most prevalent tumour forms globally. In 2018, melanoma was predicted to be the sixteenth most common cancer in humans, with 300,000 new cases. The recommended strategy outperforms current state-of-the-art methods to skin cancer detection, according to results from a simulated comparison.