Skip to content

Using CNNs and Sentinel-2 satellite data to predict landslides

Notifications You must be signed in to change notification settings

whigg/natural_disaster_pred

 
 

Repository files navigation

Predicting Landslides using CNNs and Sentinel-2 Data

Quick outline of files:

Landslide_detection_main.py trains the network and does most of the other work, uses create_dataset.py to create a dataset out of images in a folder called 'earth_engine_good'.

Accuracy_plotter.py creates a plot of accuracies during training based on the text files in results folder(copy pasted from what the python file prints).

image_download_script.js can be pasted directly into the https://code.earthengine.google.com console and you just have to enter the coordinates of the earthquake into cropping_coordinates.py to get the bounding boxes.

N.B.: To run cross validation train on different values for eval_sessions Change the variable from [i for i in range(0,4)] to [i for i in range(4,8)] and so on for 8,12 ->12,16->16,20

About

Using CNNs and Sentinel-2 satellite data to predict landslides

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Python 92.6%
  • JavaScript 7.4%