Skin Cancer Detection: CNN vs SVM

Collaborators: Alessio Desideri

The Project

Starting from over 27,000 dermoscopic images of moles and skin lesions sourced from the ISIC archive, the aim of this study is to compare the performance of two models — a Convolutional Neural Network (CNN) and a Support Vector Machine (SVM) — on a binary classification task (malignant cancer: yes/no).

Visit the archive: https://gallery.isic-archive.com/#!/topWithHeader/onlyHeaderTop/gallery?filter=%5B%5D

Repository Contents

• Final Report (PDF): The comprehensive write-up of our methodology, experiments, results and conclusions.
• Jupyter Notebook: All the code used for data preprocessing, model training, evaluation and visualization.