Implementation of Convolutional Neural Networks in Skin Cancer Classification Using Django

EasyChair Preprint 3830

Abstract

Skin diseases can be clearly seen by yourself or others. Although this disease is clearly visible on the skin, sometimes we are very worried, for example if this skin disease is not a mild disease. So there are some people if you experience skin diseases directly and will quickly go to a dermatologist, both to check complaints and symptoms experienced. This skin is very protective of the body especially from the sun, so that if something unexpected happens it can result in death. One example of a deadly skin disease is skin cancer or cancerous tumor. In this study the classification of skin cancer will be Banign and Malignant with CNN algorithm. Where the dataset used was 3297 skin cancer images taken from the Kaggle website. We use 2 different CNN architectures. The first architecture has 6,427,745 parameters, and the second is 2,797,665. With these two architectures, the accuracy values ​​range from 70-90%, the first model has an accuracy value of 93%, and the second model has 74%. We did training for many times, each time we did 10 epoches of repetition, and every epoch of 100-200 iterations.

Keyphrases: CNN, Django, Malignant, banign, scin cancer

@booklet{EasyChair:3830,
  author    = {Nur Nafi'Iyah},
  title     = {Implementation of Convolutional Neural Networks in Skin Cancer Classification Using Django},
  howpublished = {EasyChair Preprint 3830},
  year      = {EasyChair, 2020}}