Proceedings of the First International Conference on Advanced Scientific Innovation in Science, Engineering and Technology, ICASISET 2020, 16-17 May 2020, Chennai, India

Research Article

Skin Cancer Recognition and Detection Using Machine Learning Algorithm

Download435 downloads
  • @INPROCEEDINGS{10.4108/eai.16-5-2020.2304046,
        author={A  Jenitha and G  Amrutha and K. L.  Kishore and K.R.  Rohan and S. N.  Sagar},
        title={Skin Cancer Recognition and Detection Using Machine Learning Algorithm},
        proceedings={Proceedings of the First  International Conference on Advanced Scientific Innovation in Science, Engineering and Technology, ICASISET 2020, 16-17 May 2020, Chennai, India},
        publisher={EAI},
        proceedings_a={ICASISET},
        year={2021},
        month={1},
        keywords={skin cancer cnn melanoma svm knn machine learning},
        doi={10.4108/eai.16-5-2020.2304046}
    }
    
  • A Jenitha
    G Amrutha
    K. L. Kishore
    K.R. Rohan
    S. N. Sagar
    Year: 2021
    Skin Cancer Recognition and Detection Using Machine Learning Algorithm
    ICASISET
    EAI
    DOI: 10.4108/eai.16-5-2020.2304046
A Jenitha1,*, G Amrutha1, K. L. Kishore1, K.R. Rohan1, S. N. Sagar1
  • 1: Dr. T. Thimmaiah Institute of Technology Visveswaraya Technological University KGF, Karnataka India
*Contact email: jenitha@drttit.edu.in

Abstract

In this paper, we concentrate on the identification of skin cancer. The skin images are taken from a medical database which is a pre-processed image, which is given as input for different machine learning algorithm. The algorithm used is KNN classifier, SVM classifier, and CNN model. where these classifiers will classify whether a given image is cancerous or non-cancerous image. In case of the KNN and SVM the output is 80%, hence in CNN model substantial improvement in accuracy of cancer detection is obtained & it can classify the cancerous & Non-cancerous images efficiently. The process was conducted for test data, training data and validation data using different-images. The training dataset was trained with 100 epochs. The process obtained the accuracy of 97% in training result. in testing result obtained is 95% of accuracy and 96% for val-idation testing.