model-2.jpg
MIDS Capstone Project Fall 2019

RoadScanner: Resist efficiency with beauty!

RoadScanner enables travelers to enhance their travel experience and offers an alternative perspective to trip planning. 

While all existing trip planning tools focus on quantitative trip characteristics like duration, distance, time or cost, RoadScanner's focal point is qualitative and leverages state of the art machine learning techniques to predict how scenic a trip will be. RoadScanner takes into consideration the scenery variation for different seasons or daylight conditions, and even offers a preview of the predicted trip vista, customised for the user preferences.

RoadScanner is..

  • Using a highly accurate Image Classification model to score any route and predict the view based on the user trip preferences.
  • Leveraged the 360° panoramic view of Google Street View images to train the model.
  • Using CycleGANs to generate the view of alternative trip conditions.

 

model-1.jpg

"If the destination is heaven, why do we scramble to be first in line for hell?"

— Douglas Horton

Video

Last updated:

December 17, 2019