MSc thesis subject: ScenicNet: understanding scenicness with deep learning

What makes a place beautiful? What makes us love a landscape? Understanding what makes a landscape beautiful could help building new tourist routes, or, in any case, understand if a landscape is beautiful per se or only because we photograph it in a certain way.

Automated vision tasks, such as the classification of objects in still images, have seen tremendous progress in the respective research fields of Computer Vision and Machine Learning. This can be traced back to the introduction of Convolutional Neural Networks (CNNs), a family of powerful and versatile machine learning algorithms. They work by being trained on extremely large, annotated datasets that contain natural images of everyday objects. The theoretical concept of CNNs enables us to also use them for predicting all kinds of quantities, provided that we have a large dataset of input (images) and output (the scenicness) pairs. We will use a dataset built in Great Britain providing exactly that to undertand beauty of landscapes in the UK.


  • Develop a CNN model able to predict scenicness of a landscape based on users pictures
  • Enrich the model with inputs from remote sensing (e.g. aerial or Sentinel 2 images)
  • Use the model for prediction in controlled scenarios, to understand the role of external factors in scenicness (e.g. camera settings)


  • Krizhevsky, Alex, Ilya Sutskever, and Geoffrey E. Hinton. "Imagenet classification with deep convolutional neural networks." Advances in neural information processing systems. 2012.
  • Workman, Scott, Souvenir, Richard, Jacobs, Nathan. “Understanding and Mapping Natural Beauty“. International conference on computer vision. 2017


  • Programming skills in MATLAB or Python (or high motivation for learning);
  • Some background in statistics and/or machine learning is an asset.

Theme(s): Modelling & visualisation, Human – space interaction