EuroSAT classification task

Intro

This repository contains my work for the class "Deep Learning for Image and Video Processing" taught at Maastricht University by professor Alexia Briassouli.

Author: Bianca M. Massacci
Student Number: i6261625
Professor: Alexia Briassouli
Class: Deep Learning for Image and Video Processing - KEN4244 \

Master in Data Science for Decision Making.
Department of Advanced Computing Sciences (DACS).
Faculty of Science and Engineering

Directory Structure and File Contents

Because of GPU requirements, I wrote and ran all my scripts on Kaggle. Follows a brief description of each script's content and purpose. In general, the scripts share the data reading and preprocessing segments, as well as plots and evaluation metrics. The main difference between scripts lays in the architecture used.

• "EuroSAT_classification/VGG16.ipynb"
• "EuroSAT_classification/eurosat-ablation-vgg16.ipynb"
• "EuroSAT_classification/eurosat-resnet50.ipynb"
• "EuroSAT_classification/eurosat-vgg16-not-pretrained.ipynb"
• "EuroSAT_classification/eurosat-vgg19.ipynb"
• "EuroSAT_classification/preprocessing.ipynb"

References and Credits

• Kaggle for the space and the GPU
• The original Kaggle notebook for the VGG16 idea and general data reading and script set-up