/sat-6-classification

Notebooks for exploring and modeling SAT-6 satellite imagery data

Primary LanguageJupyter NotebookMIT LicenseMIT

SAT-6 Classification Project

The goal of this project is to build robust classifiers for the SAT-6 dataset. There are 4 notebooks in the project demonstrate the process for creating the classifiers:

  1. eda.ipynb: The exploratory data analysis notebook. Here are the data exploration notes, visualizations, and a simple baseline model.
  2. cnn.ipynb: A deep convolutional neural network trained on SAT-6.
  3. cnn_eval.ipynb: An exploration of the neural network trained in cnn.ipynb.
  4. vae.ipynb: A variational autoencoder trained on SAT-6.
  5. vae_eval: An exploration of the autoencoder and its resulting vector space created in vae.ipynb.