Training a CNN to classify images of different Landscapes

Original article was published on Deep Learning on Medium


Training a CNN to classify images of different Landscapes

Library used : PyTorch

The data was taken from the INTEL image classification dataset.

Photo by Pietro De Grandi on Unsplash

Welcome to this quick read on how to use Transfer Learning to classify various landscape images like the one you see above.

The dataset can be found on the Kaggle website, link : https://www.kaggle.com/puneet6060/intel-image-classification

The dataset contains three sub-folders:

  • seg_train : The training data that our model will use to learn.
  • seg_test : The validation data to test our model’s accuracy.
  • seg_pred : The data on which we are to make our predictions.

We are to extract these data and store them into datasets, in order to perform some improvements before converting them to datasets,

We use Data-Augmentation :