Detection of Melanoma Using Deep Learning Techniques
R Hamsalekha, G. Glan Devadhas, T Y Satheesha · 2024
Being a very destructive type of skin cancer, melanoma requires early discovery in order to effectively treat it. If it is not detected early the rate of survival is very less. Recently there are developments in Deep learning techniques, which have shown improved performance in a various types of computer vision applications, especially with regard to Convolutional Neural Networks (CNNs) and Vision Transformers (ViTs) In order to improve the identification of melanoma in dermoscopic images, this research suggests a hybrid model that combines the advantages of Vision Transformers and CNNs. The suggested approach makes use of a CNN foundation as framework to extract local features from pictures of skin lesions, and then a Vision Transformer module which helps to obtain global context and long-term dependencies. By Comparing a hybrid model to conventional CNNs and independent transformer models.