/Leonardo-Labs-Competition

A private kaggle competition organized by Leonardo Labs and Sapienza University of Rome.

Primary LanguageJupyter Notebook

Kaggle-Competition

Image Classification with Real World Data Distributions

Schermata 2022-09-09 alle 15 06 12

This private Kaggle competition was hosted by Leonardo Labs: https://www.leonardo.com/en/innovation-technology/leonardo-labs. In this competition we had to classify 7 different classes of marine ships:

  1. battleships
  2. coast-guard
  3. containerships
  4. cruise-ships
  5. drilling-rigs
  6. motor-yachts
  7. submarines

We trained an Artificial Neural Network to carry out a multiclass image classification task. We obtained an accuracy ~95%.

What's in here?

  • The code we developed is in the KaggleCompetitionFinal.ipynb file,
  • The training and testing images in the associated folder
  • The results obtained can be found in the file submission.csv where are reported the names of the images tested and the predicted class number.

Dataset

The dataset was provided via the Kaggle platform and is available in the images folder of this repository.

Disclamair

This project was part of a training camp organize by promoted by Leonardo Labs and Prof. F. Galasso at Sapienza University of Rome. This project was done in collaboration with Andrea Conti, Hassan Raza and Simone Boesso.