semantic_segmentation_satellite_image

Basic Information:

In this project, I develop machine learning models to detect geological features from Landsat 8 images. In this project, I aim to develop machine learning models that can detect different geologic features especially faults from satellite images. I have used a deep neural network and random forest algorithms to identify features from Landsat 8 images. The model is powerful that it can even detect recently developed river bar, and sharp fault bends. I am still working on the model to make it better predictor for geoscientists.

For inquiry:

For anything about the implementations, please feel free to write me an email :

Sabber Ahamed sabbers@gmail.com Github: msahamed medium: @sabber LinkedIn: sabber-ahamed

Notebook files/folder :

image_classification.ipynb : Image processing and Model building

shape_files : Shape files of the features (river, elevated region, river bar, low land)

Codes and libraries

This project requires Python 3+. I have Used python 3.6.4. The following Python libraries are also required:

NumPy - Numerical tool

Pandas - Dataframe

matplotlib - Plotting library

scikit-learn - Classical machine leaning library

Keras - Deep learning library

Gdal - GIS and remote sensing data processing tool

Datasets

Landsat Images can be downloaded from USGS earth explorer

Bug reports

Bug reports, comments, and suggestions are always welcome. The best the channel is to create an issue on the Issue Tracker here at the repository : https://github.com/msahamed/semantic_segmentation_satellite_image/blob/master/image_classification.ipynb

License

This program is free software: you can redistribute it and modify it under the terms of the MIT.