Skin cancer classification on HAM10000

This repository contains the project made for the course of Cognitive Services at the Master degree in computer science at the University of Padua.

In report/ folder there is the paper associated to the project. It explains the choices, the approach and the dataset that we used in the project.

In code/ folder there is the code to reproduce the experiments performed for the project and explained in the project report:

• model_building.py creates the five CNN models explained in the Proposed model section of the report;
• model_testing.py computes the metrics scores associated with the five models built by the previous script;
• learning_rate_test.py trains the proposed model with the three learning rate values explained in the Optimizer and optimizer hyperparameters choice section of the report;
• data_augmentation.py creates the oversampled dataset to perform the experiment explained in the Experiments section of the report;
• model.py performs all the experiments presented in the Experiments section of the report;
• model_test.py computes the metrics scores associated with the models built by the previous script.

In results/ folder there are all the results obtained in this project and showed in the project report.

In presentationSlides/ folder there are the slides used to present this project.

In courseSlides/ folder there are the slides used by Professors to explain the course.

In usefullInfo/ folder there are some useful informations about skin cancer classification task and deep learning topics.