Project Icey Ships: Identification of ships or icebergs from satellite images

Project Information

  • Partner: CSIR
  • Students: Khutsiso Matli (UL), Themba Mathebula (SMU) and Thabang Mautjane (UL)
  • Project Lead: Dhiren Seethram
  • Project Mentors: Rory Meyer and Themba Mathebula
  • Year: 2017/2018

Project Description

Project Icey-Ships is project from the Statoil/C-CORE competitions from the Kaggle website. The Icey-Ships ‘s main objective is to create a classifier that will classify or detect whether the Synthetic Aperture Radar (SAR) image contain a ship or an iceberg.The Icey-Ship team collected the train and test dataset from the kaggle website and used it for analysis, we cleaned the data that was loaded from JSON(JavaScript Object Notation) file and converted it to CSV(Comma Separated Values) file using jupyter notebook. Furthermore, the Icey-Ships team created a classifier model using convulutional neuron network(CNN). The model will be used to classify whether an image contain a ship or an iceberg.