dzubke / oil_well_detector

A machine learning algorithm that is meant to detect oil wells in the Bakken oil field based on satellite imagery

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

oil-well-detection

Detections oil wells in the Bakken oil field based on satellite imagery

For this project, I will be training an image recognition algorithm to recognize if a pixel is part of an oil well. I will be analyzing various types of satellite images to conduct this analysis.

I am still working on this project. So far I have:

  • downloaded the relevant satelittle images (the raw images are too big to be in a github repo)
  • utilized Rasterio and Shapely analyze the features of the image and crop an area of interest

Still to come I need to:

  • Create the training data
    • make outline examples of oil wells in polygons using Google Earth
    • convert those polygon outlines into a raster image with positive values (1) inside the polygon and zero values outside of the polygon
  • write the neural network that will be trained on the training data
  • compare the performance of a full-connected neural network with a convolutional neural network in predicting if a given pixel will be part of an oil well.

About

A machine learning algorithm that is meant to detect oil wells in the Bakken oil field based on satellite imagery

License:MIT License


Languages

Language:Jupyter Notebook 99.1%Language:Python 0.9%